Real-time Domain Adaptation in Semantic Segmentation.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "o3jrB7Oq8FVf",
        "outputId": "886e0e58-bff3-4de7-bb02-05e3cc98e31e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting tensorboardX==1.4\n",
            "  Downloading tensorboardX-1.4-py2.py3-none-any.whl (67 kB)\n",
            "\u001b[?25l\r\u001b[K     |████▉                           | 10 kB 31.9 MB/s eta 0:00:01\r\u001b[K     |█████████▊                      | 20 kB 9.2 MB/s eta 0:00:01\r\u001b[K     |██████████████▋                 | 30 kB 8.3 MB/s eta 0:00:01\r\u001b[K     |███████████████████▌            | 40 kB 7.9 MB/s eta 0:00:01\r\u001b[K     |████████████████████████▍       | 51 kB 4.8 MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████▎  | 61 kB 5.3 MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 67 kB 2.8 MB/s \n",
            "\u001b[?25hRequirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from tensorboardX==1.4) (1.15.0)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from tensorboardX==1.4) (1.19.5)\n",
            "Requirement already satisfied: protobuf>=3.2.0 in /usr/local/lib/python3.7/dist-packages (from tensorboardX==1.4) (3.17.3)\n",
            "Installing collected packages: tensorboardX\n",
            "Successfully installed tensorboardX-1.4\n"
          ]
        }
      ],
      "source": [
        "!pip install tensorboardX==1.4"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mEi2CsDq--2b",
        "outputId": "c779db61-a4db-4230-ef26-86d393f359ed"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "tmA_KI_u_Dib"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import logging"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "18K55-maFLqT",
        "outputId": "d6e90413-fe41-473e-f06d-f32abf4aa45a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Cloning into 'AMLexam'...\n",
            "remote: Enumerating objects: 282, done.\u001b[K\n",
            "remote: Counting objects: 100% (282/282), done.\u001b[K\n",
            "remote: Compressing objects: 100% (213/213), done.\u001b[K\n",
            "remote: Total 282 (delta 123), reused 179 (delta 55), pack-reused 0\u001b[K\n",
            "Receiving objects: 100% (282/282), 329.25 KiB | 5.88 MiB/s, done.\n",
            "Resolving deltas: 100% (123/123), done.\n"
          ]
        }
      ],
      "source": [
        "if not os.path.isdir('./AMLexam'):\n",
        "\n",
        "  !git clone 'https://github.com/frarus/AMLexam.git'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Etr9-Z1NMWzZ",
        "outputId": "d9317548-298c-4120-a4dc-57af4722e998"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "/content/AMLexam\n"
          ]
        }
      ],
      "source": [
        "cd AMLexam/"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Vt6bstMkMdYI",
        "outputId": "4a9ffd11-666e-4b6b-e35b-4243ee098cfd"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "/content/AMLexam/BiSeNet\n"
          ]
        }
      ],
      "source": [
        "cd BiSeNet/"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        },
        "id": "CFgMCP9BC0Kg",
        "outputId": "effd2a44-d5f1-46f6-da3a-ec5630cbefcd"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# GET INFO ABOUT LABELS AND THEIR COLORS TO PRINT A LEGEND\n",
        "# info.json must be on Google Drive in the folder Datasets/Cityscapes\n",
        "import json\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "infojson = json.load(open('/content/drive/MyDrive/Datasets/Cityscapes/info.json', 'r'))\n",
        "colors=infojson['palette']\n",
        "classes=infojson['label']\n",
        "f=lambda m,c: plt.plot([],[], marker=m, color=c, ls=\"none\")[0]\n",
        "colors=np.array(np.array(colors)/255, dtype=np.float32)\n",
        "handles=[f(\"s\", colors[i]) for i in range(20)]\n",
        "legend=plt.legend(handles, classes, loc=3, frameon=False)\n",
        "fig = legend.figure\n",
        "fig.canvas.draw()\n",
        "bbox=legend.get_window_extent().transformed(fig.dpi_scale_trans.inverted())\n",
        "plt.gca().set_axis_off()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fO7p9Xw6FDyD"
      },
      "source": [
        "STEP 2"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aXWiYJ-okJw2",
        "outputId": "f687f122-45ed-42e9-a4a8-4bd81c5bc2b9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/resnet18-f37072fd.pth\" to /root/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth\n",
            "100% 44.7M/44.7M [00:00<00:00, 172MB/s]\n",
            "Downloading: \"https://download.pytorch.org/models/resnet101-63fe2227.pth\" to /root/.cache/torch/hub/checkpoints/resnet101-63fe2227.pth\n",
            "100% 171M/171M [00:00<00:00, 184MB/s]\n",
            "epoch 0, lr 0.025000: 100% 500/500 [02:02<00:00,  4.09it/s, loss=1.327232]\n",
            "loss for train : 2.407238\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.239\n",
            "epoch 1, lr 0.024550: 100% 500/500 [01:11<00:00,  7.04it/s, loss=1.335859]\n",
            "loss for train : 1.306158\n",
            "start val!\n",
            "precision per pixel for test: 0.761\n",
            "mIoU for validation: 0.315\n",
            "epoch 2, lr 0.024098: 100% 500/500 [01:11<00:00,  7.00it/s, loss=1.069794]\n",
            "loss for train : 1.060273\n",
            "start val!\n",
            "precision per pixel for test: 0.744\n",
            "mIoU for validation: 0.329\n",
            "epoch 3, lr 0.023646: 100% 500/500 [01:11<00:00,  6.97it/s, loss=0.706030]\n",
            "loss for train : 0.888361\n",
            "start val!\n",
            "precision per pixel for test: 0.738\n",
            "mIoU for validation: 0.343\n",
            "epoch 4, lr 0.023193: 100% 500/500 [01:11<00:00,  6.97it/s, loss=0.656190]\n",
            "loss for train : 0.831048\n",
            "start val!\n",
            "precision per pixel for test: 0.754\n",
            "mIoU for validation: 0.360\n",
            "epoch 5, lr 0.022738: 100% 500/500 [01:11<00:00,  6.99it/s, loss=0.678417]\n",
            "loss for train : 0.738839\n",
            "start val!\n",
            "precision per pixel for test: 0.782\n",
            "mIoU for validation: 0.399\n",
            "epoch 6, lr 0.022283: 100% 500/500 [01:11<00:00,  6.96it/s, loss=0.485738]\n",
            "loss for train : 0.629797\n",
            "start val!\n",
            "precision per pixel for test: 0.784\n",
            "mIoU for validation: 0.442\n",
            "epoch 7, lr 0.021827: 100% 500/500 [01:11<00:00,  6.98it/s, loss=0.556276]\n",
            "loss for train : 0.574872\n",
            "start val!\n",
            "precision per pixel for test: 0.780\n",
            "mIoU for validation: 0.460\n",
            "epoch 8, lr 0.021369: 100% 500/500 [01:11<00:00,  7.00it/s, loss=0.683338]\n",
            "loss for train : 0.505108\n",
            "start val!\n",
            "precision per pixel for test: 0.797\n",
            "mIoU for validation: 0.488\n",
            "epoch 9, lr 0.020911: 100% 500/500 [01:13<00:00,  6.76it/s, loss=0.561551]\n",
            "loss for train : 0.451690\n",
            "start val!\n",
            "precision per pixel for test: 0.800\n",
            "mIoU for validation: 0.510\n",
            "epoch 10, lr 0.020451: 100% 500/500 [01:14<00:00,  6.75it/s, loss=0.564219]\n",
            "loss for train : 0.427096\n",
            "start val!\n",
            "precision per pixel for test: 0.796\n",
            "mIoU for validation: 0.494\n",
            "epoch 11, lr 0.019991: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.405920]\n",
            "loss for train : 0.403193\n",
            "start val!\n",
            "precision per pixel for test: 0.803\n",
            "mIoU for validation: 0.529\n",
            "epoch 12, lr 0.019529: 100% 500/500 [01:12<00:00,  6.94it/s, loss=0.339734]\n",
            "loss for train : 0.387163\n",
            "start val!\n",
            "precision per pixel for test: 0.801\n",
            "mIoU for validation: 0.525\n",
            "epoch 13, lr 0.019066: 100% 500/500 [01:11<00:00,  6.98it/s, loss=0.374762]\n",
            "loss for train : 0.362339\n",
            "start val!\n",
            "precision per pixel for test: 0.804\n",
            "mIoU for validation: 0.537\n",
            "epoch 14, lr 0.018601: 100% 500/500 [01:11<00:00,  6.98it/s, loss=0.326455]\n",
            "loss for train : 0.356505\n",
            "start val!\n",
            "precision per pixel for test: 0.798\n",
            "mIoU for validation: 0.520\n",
            "epoch 15, lr 0.018135: 100% 500/500 [01:11<00:00,  6.97it/s, loss=0.491292]\n",
            "loss for train : 0.401928\n",
            "start val!\n",
            "precision per pixel for test: 0.795\n",
            "mIoU for validation: 0.484\n",
            "epoch 16, lr 0.017668: 100% 500/500 [01:11<00:00,  6.98it/s, loss=0.379394]\n",
            "loss for train : 0.373880\n",
            "start val!\n",
            "precision per pixel for test: 0.799\n",
            "mIoU for validation: 0.515\n",
            "epoch 17, lr 0.017200: 100% 500/500 [01:11<00:00,  6.99it/s, loss=0.328819]\n",
            "loss for train : 0.331737\n",
            "start val!\n",
            "precision per pixel for test: 0.802\n",
            "mIoU for validation: 0.525\n",
            "epoch 18, lr 0.016730: 100% 500/500 [01:11<00:00,  6.97it/s, loss=0.335851]\n",
            "loss for train : 0.313675\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.536\n",
            "epoch 19, lr 0.016259: 100% 500/500 [01:11<00:00,  6.99it/s, loss=0.350315]\n",
            "loss for train : 0.302953\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.532\n",
            "epoch 20, lr 0.015786: 100% 500/500 [01:11<00:00,  7.00it/s, loss=0.276671]\n",
            "loss for train : 0.299998\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.542\n",
            "epoch 21, lr 0.015312: 100% 500/500 [01:11<00:00,  7.01it/s, loss=0.337947]\n",
            "loss for train : 0.288432\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.538\n",
            "epoch 22, lr 0.014836: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.217170]\n",
            "loss for train : 0.282319\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.542\n",
            "epoch 23, lr 0.014358: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.242317]\n",
            "loss for train : 0.278442\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.535\n",
            "epoch 24, lr 0.013879: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.240067]\n",
            "loss for train : 0.271801\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.537\n",
            "epoch 25, lr 0.013397: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.209908]\n",
            "loss for train : 0.267244\n",
            "start val!\n",
            "precision per pixel for test: 0.804\n",
            "mIoU for validation: 0.539\n",
            "epoch 26, lr 0.012914: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.279543]\n",
            "loss for train : 0.262653\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.538\n",
            "epoch 27, lr 0.012429: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.301635]\n",
            "loss for train : 0.258979\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.543\n",
            "epoch 28, lr 0.011941: 100% 500/500 [01:14<00:00,  6.75it/s, loss=0.218349]\n",
            "loss for train : 0.255061\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.547\n",
            "epoch 29, lr 0.011452: 100% 500/500 [01:14<00:00,  6.75it/s, loss=0.222431]\n",
            "loss for train : 0.256072\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.532\n",
            "epoch 30, lr 0.010960: 100% 500/500 [01:13<00:00,  6.76it/s, loss=0.310820]\n",
            "loss for train : 0.254271\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.542\n",
            "epoch 31, lr 0.010465: 100% 500/500 [01:14<00:00,  6.75it/s, loss=0.191385]\n",
            "loss for train : 0.247753\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.544\n",
            "epoch 32, lr 0.009968: 100% 500/500 [01:13<00:00,  6.76it/s, loss=0.292155]\n",
            "loss for train : 0.243748\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.544\n",
            "epoch 33, lr 0.009468: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.242870]\n",
            "loss for train : 0.240756\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.546\n",
            "epoch 34, lr 0.008966: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.228853]\n",
            "loss for train : 0.239506\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.547\n",
            "epoch 35, lr 0.008460: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.292219]\n",
            "loss for train : 0.236198\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.546\n",
            "epoch 36, lr 0.007950: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.220048]\n",
            "loss for train : 0.234755\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.543\n",
            "epoch 37, lr 0.007437: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.236179]\n",
            "loss for train : 0.232914\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.543\n",
            "epoch 38, lr 0.006920: 100% 500/500 [01:13<00:00,  6.76it/s, loss=0.269681]\n",
            "loss for train : 0.231160\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.545\n",
            "epoch 39, lr 0.006399: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.234353]\n",
            "loss for train : 0.229403\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.542\n",
            "epoch 40, lr 0.005873: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.269647]\n",
            "loss for train : 0.227654\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.543\n",
            "epoch 41, lr 0.005342: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.219886]\n",
            "loss for train : 0.225829\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.546\n",
            "epoch 42, lr 0.004804: 100% 500/500 [01:13<00:00,  6.76it/s, loss=0.271232]\n",
            "loss for train : 0.225192\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.540\n",
            "epoch 43, lr 0.004260: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.284724]\n",
            "loss for train : 0.222375\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.540\n",
            "epoch 44, lr 0.003709: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.159093]\n",
            "loss for train : 0.222085\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.541\n",
            "epoch 45, lr 0.003147: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.238100]\n",
            "loss for train : 0.221030\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.541\n",
            "epoch 46, lr 0.002575: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.190737]\n",
            "loss for train : 0.219926\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.541\n",
            "epoch 47, lr 0.001987: 100% 500/500 [01:13<00:00,  6.77it/s, loss=0.268174]\n",
            "loss for train : 0.218785\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.543\n",
            "epoch 48, lr 0.001380: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.194972]\n",
            "loss for train : 0.218322\n",
            "start val!\n",
            "precision per pixel for test: 0.805\n",
            "mIoU for validation: 0.543\n",
            "epoch 49, lr 0.000739: 100% 500/500 [01:13<00:00,  6.78it/s, loss=0.286894]\n",
            "loss for train : 0.218202\n",
            "start val!\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.541\n"
          ]
        }
      ],
      "source": [
        "#RESNET 101\n",
        "!python train.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "otZXxzsIlCi1",
        "outputId": "53dc9092-2480-4d27-ac17-2c76fe33695e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd/best_crossentropy_loss.pth ...\n",
            "The best epoch is 28\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:42<00:00,  1.37s/it]\n",
            "IoU for class road: 0.965\n",
            "IoU for class sidewalk: 0.737\n",
            "IoU for class building: 0.864\n",
            "IoU for class wall: 0.335\n",
            "IoU for class fence: 0.272\n",
            "IoU for class pole: 0.377\n",
            "IoU for class light: 0.401\n",
            "IoU for class sign: 0.548\n",
            "IoU for class vegetation: 0.879\n",
            "IoU for class terrain: 0.515\n",
            "IoU for class sky: 0.897\n",
            "IoU for class person: 0.618\n",
            "IoU for class rider: 0.359\n",
            "IoU for class car: 0.887\n",
            "IoU for class truck: 0.299\n",
            "IoU for class bus: 0.403\n",
            "IoU for class train: 0.143\n",
            "IoU for class motocycle: 0.313\n",
            "IoU for class bicycle: 0.584\n",
            "precision per pixel for test: 0.806\n",
            "mIoU for validation: 0.547\n"
          ]
        }
      ],
      "source": [
        "#RESNET 101\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fYaTBiTZwR4Q",
        "outputId": "80efa0d1-d5d1-400a-b00e-12bf3f1794c2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch 0, lr 0.025000: 100% 500/500 [00:38<00:00, 13.03it/s, loss=1.407807]\n",
            "loss for train : 2.793031\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.220\n",
            "epoch 1, lr 0.024550: 100% 500/500 [00:38<00:00, 12.93it/s, loss=1.741456]\n",
            "loss for train : 1.669413\n",
            "start val!\n",
            "precision per pixel for test: 0.731\n",
            "mIoU for validation: 0.261\n",
            "epoch 2, lr 0.024098: 100% 500/500 [00:39<00:00, 12.70it/s, loss=1.080343]\n",
            "loss for train : 1.375189\n",
            "start val!\n",
            "precision per pixel for test: 0.750\n",
            "mIoU for validation: 0.291\n",
            "epoch 3, lr 0.023646: 100% 500/500 [00:39<00:00, 12.51it/s, loss=1.209980]\n",
            "loss for train : 1.202755\n",
            "start val!\n",
            "precision per pixel for test: 0.762\n",
            "mIoU for validation: 0.328\n",
            "epoch 4, lr 0.023193: 100% 500/500 [00:39<00:00, 12.76it/s, loss=0.876690]\n",
            "loss for train : 1.055839\n",
            "start val!\n",
            "precision per pixel for test: 0.757\n",
            "mIoU for validation: 0.342\n",
            "epoch 5, lr 0.022738: 100% 500/500 [00:39<00:00, 12.76it/s, loss=0.812608]\n",
            "loss for train : 0.947127\n",
            "start val!\n",
            "precision per pixel for test: 0.769\n",
            "mIoU for validation: 0.364\n",
            "epoch 6, lr 0.022283: 100% 500/500 [00:39<00:00, 12.73it/s, loss=0.916740]\n",
            "loss for train : 0.892096\n",
            "start val!\n",
            "precision per pixel for test: 0.776\n",
            "mIoU for validation: 0.400\n",
            "epoch 7, lr 0.021827: 100% 500/500 [00:38<00:00, 12.83it/s, loss=0.749314]\n",
            "loss for train : 0.794918\n",
            "start val!\n",
            "precision per pixel for test: 0.780\n",
            "mIoU for validation: 0.424\n",
            "epoch 8, lr 0.021369: 100% 500/500 [00:39<00:00, 12.75it/s, loss=0.520558]\n",
            "loss for train : 0.719665\n",
            "start val!\n",
            "precision per pixel for test: 0.778\n",
            "mIoU for validation: 0.423\n",
            "epoch 9, lr 0.020911: 100% 500/500 [00:38<00:00, 12.83it/s, loss=0.895841]\n",
            "loss for train : 0.681391\n",
            "start val!\n",
            "precision per pixel for test: 0.773\n",
            "mIoU for validation: 0.403\n",
            "epoch 10, lr 0.020451: 100% 500/500 [00:38<00:00, 12.85it/s, loss=0.552894]\n",
            "loss for train : 0.636218\n",
            "start val!\n",
            "precision per pixel for test: 0.787\n",
            "mIoU for validation: 0.446\n",
            "epoch 11, lr 0.019991: 100% 500/500 [00:39<00:00, 12.80it/s, loss=0.694907]\n",
            "loss for train : 0.602246\n",
            "start val!\n",
            "precision per pixel for test: 0.788\n",
            "mIoU for validation: 0.450\n",
            "epoch 12, lr 0.019529: 100% 500/500 [00:39<00:00, 12.78it/s, loss=0.540886]\n",
            "loss for train : 0.576766\n",
            "start val!\n",
            "precision per pixel for test: 0.788\n",
            "mIoU for validation: 0.466\n",
            "epoch 13, lr 0.019066: 100% 500/500 [00:38<00:00, 12.86it/s, loss=0.487733]\n",
            "loss for train : 0.551351\n",
            "start val!\n",
            "precision per pixel for test: 0.791\n",
            "mIoU for validation: 0.460\n",
            "epoch 14, lr 0.018601: 100% 500/500 [00:38<00:00, 12.87it/s, loss=0.494105]\n",
            "loss for train : 0.523206\n",
            "start val!\n",
            "precision per pixel for test: 0.789\n",
            "mIoU for validation: 0.455\n",
            "epoch 15, lr 0.018135: 100% 500/500 [00:39<00:00, 12.82it/s, loss=0.668522]\n",
            "loss for train : 0.516596\n",
            "start val!\n",
            "precision per pixel for test: 0.790\n",
            "mIoU for validation: 0.467\n",
            "epoch 16, lr 0.017668: 100% 500/500 [00:39<00:00, 12.81it/s, loss=0.601787]\n",
            "loss for train : 0.495418\n",
            "start val!\n",
            "precision per pixel for test: 0.789\n",
            "mIoU for validation: 0.455\n",
            "epoch 17, lr 0.017200: 100% 500/500 [00:38<00:00, 13.10it/s, loss=0.323766]\n",
            "loss for train : 0.475816\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.473\n",
            "epoch 18, lr 0.016730: 100% 500/500 [00:38<00:00, 13.12it/s, loss=0.487785]\n",
            "loss for train : 0.457622\n",
            "start val!\n",
            "precision per pixel for test: 0.792\n",
            "mIoU for validation: 0.471\n",
            "epoch 19, lr 0.016259: 100% 500/500 [00:38<00:00, 13.12it/s, loss=0.472945]\n",
            "loss for train : 0.448197\n",
            "start val!\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.483\n",
            "epoch 20, lr 0.015786: 100% 500/500 [00:38<00:00, 13.10it/s, loss=0.437861]\n",
            "loss for train : 0.431190\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.486\n",
            "epoch 21, lr 0.015312: 100% 500/500 [00:38<00:00, 13.11it/s, loss=0.295537]\n",
            "loss for train : 0.418821\n",
            "start val!\n",
            "precision per pixel for test: 0.791\n",
            "mIoU for validation: 0.481\n",
            "epoch 22, lr 0.014836: 100% 500/500 [00:38<00:00, 13.13it/s, loss=0.512857]\n",
            "loss for train : 0.407338\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.476\n",
            "epoch 23, lr 0.014358: 100% 500/500 [00:37<00:00, 13.21it/s, loss=0.344468]\n",
            "loss for train : 0.403974\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.488\n",
            "epoch 24, lr 0.013879: 100% 500/500 [00:38<00:00, 13.12it/s, loss=0.326865]\n",
            "loss for train : 0.395995\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.475\n",
            "epoch 25, lr 0.013397: 100% 500/500 [00:37<00:00, 13.20it/s, loss=0.400344]\n",
            "loss for train : 0.387389\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.482\n",
            "epoch 26, lr 0.012914: 100% 500/500 [00:38<00:00, 13.06it/s, loss=0.389897]\n",
            "loss for train : 0.380422\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.481\n",
            "epoch 27, lr 0.012429: 100% 500/500 [00:37<00:00, 13.19it/s, loss=0.336369]\n",
            "loss for train : 0.373807\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.482\n",
            "epoch 28, lr 0.011941: 100% 500/500 [00:37<00:00, 13.19it/s, loss=0.330588]\n",
            "loss for train : 0.367485\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.482\n",
            "epoch 29, lr 0.011452: 100% 500/500 [00:37<00:00, 13.18it/s, loss=0.367216]\n",
            "loss for train : 0.362994\n",
            "start val!\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.488\n",
            "epoch 30, lr 0.010960: 100% 500/500 [00:38<00:00, 13.14it/s, loss=0.417670]\n",
            "loss for train : 0.358108\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.474\n",
            "epoch 31, lr 0.010465: 100% 500/500 [00:37<00:00, 13.17it/s, loss=0.415055]\n",
            "loss for train : 0.350843\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.477\n",
            "epoch 32, lr 0.009968: 100% 500/500 [00:37<00:00, 13.23it/s, loss=0.352676]\n",
            "loss for train : 0.345623\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.473\n",
            "epoch 33, lr 0.009468: 100% 500/500 [00:37<00:00, 13.16it/s, loss=0.395698]\n",
            "loss for train : 0.342402\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.477\n",
            "epoch 34, lr 0.008966: 100% 500/500 [00:37<00:00, 13.16it/s, loss=0.302579]\n",
            "loss for train : 0.337949\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.480\n",
            "epoch 35, lr 0.008460: 100% 500/500 [00:38<00:00, 13.13it/s, loss=0.448802]\n",
            "loss for train : 0.337098\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.483\n",
            "epoch 36, lr 0.007950: 100% 500/500 [00:38<00:00, 13.16it/s, loss=0.283788]\n",
            "loss for train : 0.335114\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.479\n",
            "epoch 37, lr 0.007437: 100% 500/500 [00:37<00:00, 13.17it/s, loss=0.347553]\n",
            "loss for train : 0.329795\n",
            "start val!\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.484\n",
            "epoch 38, lr 0.006920: 100% 500/500 [00:38<00:00, 13.16it/s, loss=0.364895]\n",
            "loss for train : 0.327647\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.483\n",
            "epoch 39, lr 0.006399: 100% 500/500 [00:37<00:00, 13.19it/s, loss=0.381837]\n",
            "loss for train : 0.325892\n",
            "start val!\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.479\n",
            "epoch 40, lr 0.005873: 100% 500/500 [00:38<00:00, 13.15it/s, loss=0.485380]\n",
            "loss for train : 0.322838\n",
            "start val!\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.482\n",
            "epoch 41, lr 0.005342: 100% 500/500 [00:38<00:00, 13.10it/s, loss=0.500313]\n",
            "loss for train : 0.321073\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.479\n",
            "epoch 42, lr 0.004804: 100% 500/500 [00:38<00:00, 13.12it/s, loss=0.292053]\n",
            "loss for train : 0.317885\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.477\n",
            "epoch 43, lr 0.004260: 100% 500/500 [00:38<00:00, 13.11it/s, loss=0.279916]\n",
            "loss for train : 0.315059\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.479\n",
            "epoch 44, lr 0.003709: 100% 500/500 [00:37<00:00, 13.20it/s, loss=0.351653]\n",
            "loss for train : 0.314455\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.482\n",
            "epoch 45, lr 0.003147: 100% 500/500 [00:38<00:00, 13.13it/s, loss=0.270746]\n",
            "loss for train : 0.311803\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.479\n",
            "epoch 46, lr 0.002575: 100% 500/500 [00:38<00:00, 13.15it/s, loss=0.327749]\n",
            "loss for train : 0.311456\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.478\n",
            "epoch 47, lr 0.001987: 100% 500/500 [00:38<00:00, 13.11it/s, loss=0.249806]\n",
            "loss for train : 0.308373\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.478\n",
            "epoch 48, lr 0.001380: 100% 500/500 [00:38<00:00, 13.14it/s, loss=0.253752]\n",
            "loss for train : 0.308033\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.477\n",
            "epoch 49, lr 0.000739: 100% 500/500 [00:38<00:00, 13.03it/s, loss=0.230821]\n",
            "loss for train : 0.309388\n",
            "start val!\n",
            "precision per pixel for test: 0.793\n",
            "mIoU for validation: 0.478\n"
          ]
        }
      ],
      "source": [
        "#RESNET 18\n",
        "!python train.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SGOuz2tcxeAe",
        "outputId": "714ba80e-b79a-4019-e65c-ba39a695e513"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_18_sgd/best_crossentropy_loss.pth ...\n",
            "The best epoch is 29\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:22<00:00,  1.29s/it]\n",
            "IoU for class road: 0.953\n",
            "IoU for class sidewalk: 0.663\n",
            "IoU for class building: 0.847\n",
            "IoU for class wall: 0.252\n",
            "IoU for class fence: 0.226\n",
            "IoU for class pole: 0.314\n",
            "IoU for class light: 0.309\n",
            "IoU for class sign: 0.454\n",
            "IoU for class vegetation: 0.861\n",
            "IoU for class terrain: 0.470\n",
            "IoU for class sky: 0.892\n",
            "IoU for class person: 0.555\n",
            "IoU for class rider: 0.271\n",
            "IoU for class car: 0.847\n",
            "IoU for class truck: 0.160\n",
            "IoU for class bus: 0.313\n",
            "IoU for class train: 0.096\n",
            "IoU for class motocycle: 0.257\n",
            "IoU for class bicycle: 0.535\n",
            "precision per pixel for test: 0.794\n",
            "mIoU for validation: 0.488\n"
          ]
        }
      ],
      "source": [
        "#RESNET 18\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ekq2qTJ2FH3T"
      },
      "source": [
        "STEP 3"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vjNOz75axkdx",
        "outputId": "c37f9309-2eeb-42b4-b699-2f30efe3f2b0"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting torchstat\n",
            "  Downloading torchstat-0.0.7-py3-none-any.whl (11 kB)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from torchstat) (1.19.5)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.7/dist-packages (from torchstat) (1.3.5)\n",
            "Requirement already satisfied: torch in /usr/local/lib/python3.7/dist-packages (from torchstat) (1.10.0+cu111)\n",
            "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas->torchstat) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2017.3 in /usr/local/lib/python3.7/dist-packages (from pandas->torchstat) (2018.9)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.7.3->pandas->torchstat) (1.15.0)\n",
            "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from torch->torchstat) (3.10.0.2)\n",
            "Installing collected packages: torchstat\n",
            "Successfully installed torchstat-0.0.7\n"
          ]
        }
      ],
      "source": [
        "pip install torchstat"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7lwxVZAj7HeT",
        "outputId": "ed08f94e-c3ad-4d8a-dd97-055d45c4bb08"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: AdaptiveAvgPool2d is not supported!\n",
            "[Flops]: AdaptiveAvgPool2d is not supported!\n",
            "[Memory]: AdaptiveAvgPool2d is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "[MAdd]: Sigmoid is not supported!\n",
            "[Flops]: Sigmoid is not supported!\n",
            "[Memory]: Sigmoid is not supported!\n",
            "                                       module name   input shape  output shape      params memory(MB)               MAdd             Flops   MemRead(B)  MemWrite(B) duration[%]     MemR+W(B)\n",
            "0                    saptial_path.convblock1.conv1    3 1024 512    64 512 256      1728.0      32.00      444,596,224.0     226,492,416.0    6298368.0   33554432.0       0.47%  3.985280e+07\n",
            "1                       saptial_path.convblock1.bn    64 512 256    64 512 256       128.0      32.00       33,554,432.0      16,777,216.0   33554944.0   33554432.0       0.23%  6.710938e+07\n",
            "2                     saptial_path.convblock1.relu    64 512 256    64 512 256         0.0      32.00        8,388,608.0       8,388,608.0   33554432.0   33554432.0       0.21%  6.710886e+07\n",
            "3                    saptial_path.convblock2.conv1    64 512 256   128 256 128     73728.0      16.00    4,827,643,904.0   2,415,919,104.0   33849344.0   16777216.0       1.60%  5.062656e+07\n",
            "4                       saptial_path.convblock2.bn   128 256 128   128 256 128       256.0      16.00       16,777,216.0       8,388,608.0   16778240.0   16777216.0       0.13%  3.355546e+07\n",
            "5                     saptial_path.convblock2.relu   128 256 128   128 256 128         0.0      16.00        4,194,304.0       4,194,304.0   16777216.0   16777216.0       0.12%  3.355443e+07\n",
            "6                    saptial_path.convblock3.conv1   128 256 128   256 128  64    294912.0       8.00    4,829,741,056.0   2,415,919,104.0   17956864.0    8388608.0       1.44%  2.634547e+07\n",
            "7                       saptial_path.convblock3.bn   256 128  64   256 128  64       512.0       8.00        8,388,608.0       4,194,304.0    8390656.0    8388608.0       0.09%  1.677926e+07\n",
            "8                     saptial_path.convblock3.relu   256 128  64   256 128  64         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.08%  1.677722e+07\n",
            "9                      context_path.features.conv1    3 1024 512    64 512 256      9408.0      32.00    2,457,862,144.0   1,233,125,376.0    6329088.0   33554432.0       1.02%  3.988352e+07\n",
            "10                       context_path.features.bn1    64 512 256    64 512 256       128.0      32.00       33,554,432.0      16,777,216.0   33554944.0   33554432.0       0.23%  6.710938e+07\n",
            "11                      context_path.features.relu    64 512 256    64 512 256         0.0      32.00        8,388,608.0       8,388,608.0   33554432.0   33554432.0       0.09%  6.710886e+07\n",
            "12                   context_path.features.maxpool    64 512 256    64 256 128         0.0       8.00       16,777,216.0       8,388,608.0   33554432.0    8388608.0       1.42%  4.194304e+07\n",
            "13            context_path.features.layer1.0.conv1    64 256 128    64 256 128      4096.0       8.00      266,338,304.0     134,217,728.0    8404992.0    8388608.0       0.84%  1.679360e+07\n",
            "14              context_path.features.layer1.0.bn1    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.09%  1.677773e+07\n",
            "15            context_path.features.layer1.0.conv2    64 256 128    64 256 128     36864.0       8.00    2,413,821,952.0   1,207,959,552.0    8536064.0    8388608.0       1.34%  1.692467e+07\n",
            "16              context_path.features.layer1.0.bn2    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.09%  1.677773e+07\n",
            "17            context_path.features.layer1.0.conv3    64 256 128   256 256 128     16384.0      32.00    1,065,353,216.0     536,870,912.0    8454144.0   33554432.0       0.77%  4.200858e+07\n",
            "18              context_path.features.layer1.0.bn3   256 256 128   256 256 128       512.0      32.00       33,554,432.0      16,777,216.0   33556480.0   33554432.0       0.22%  6.711091e+07\n",
            "19             context_path.features.layer1.0.relu   256 256 128   256 256 128         0.0      32.00        8,388,608.0       8,388,608.0   33554432.0   33554432.0       0.09%  6.710886e+07\n",
            "20     context_path.features.layer1.0.downsample.0    64 256 128   256 256 128     16384.0      32.00    1,065,353,216.0     536,870,912.0    8454144.0   33554432.0       0.75%  4.200858e+07\n",
            "21     context_path.features.layer1.0.downsample.1   256 256 128   256 256 128       512.0      32.00       33,554,432.0      16,777,216.0   33556480.0   33554432.0       0.21%  6.711091e+07\n",
            "22            context_path.features.layer1.1.conv1   256 256 128    64 256 128     16384.0       8.00    1,071,644,672.0     536,870,912.0   33619968.0    8388608.0       0.57%  4.200858e+07\n",
            "23              context_path.features.layer1.1.bn1    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.08%  1.677773e+07\n",
            "24            context_path.features.layer1.1.conv2    64 256 128    64 256 128     36864.0       8.00    2,413,821,952.0   1,207,959,552.0    8536064.0    8388608.0       0.68%  1.692467e+07\n",
            "25              context_path.features.layer1.1.bn2    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.08%  1.677773e+07\n",
            "26            context_path.features.layer1.1.conv3    64 256 128   256 256 128     16384.0      32.00    1,065,353,216.0     536,870,912.0    8454144.0   33554432.0       0.70%  4.200858e+07\n",
            "27              context_path.features.layer1.1.bn3   256 256 128   256 256 128       512.0      32.00       33,554,432.0      16,777,216.0   33556480.0   33554432.0       0.21%  6.711091e+07\n",
            "28             context_path.features.layer1.1.relu   256 256 128   256 256 128         0.0      32.00        8,388,608.0       8,388,608.0   33554432.0   33554432.0       0.09%  6.710886e+07\n",
            "29            context_path.features.layer1.2.conv1   256 256 128    64 256 128     16384.0       8.00    1,071,644,672.0     536,870,912.0   33619968.0    8388608.0       1.14%  4.200858e+07\n",
            "30              context_path.features.layer1.2.bn1    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.09%  1.677773e+07\n",
            "31            context_path.features.layer1.2.conv2    64 256 128    64 256 128     36864.0       8.00    2,413,821,952.0   1,207,959,552.0    8536064.0    8388608.0       1.21%  1.692467e+07\n",
            "32              context_path.features.layer1.2.bn2    64 256 128    64 256 128       128.0       8.00        8,388,608.0       4,194,304.0    8389120.0    8388608.0       0.08%  1.677773e+07\n",
            "33            context_path.features.layer1.2.conv3    64 256 128   256 256 128     16384.0      32.00    1,065,353,216.0     536,870,912.0    8454144.0   33554432.0       0.74%  4.200858e+07\n",
            "34              context_path.features.layer1.2.bn3   256 256 128   256 256 128       512.0      32.00       33,554,432.0      16,777,216.0   33556480.0   33554432.0       0.22%  6.711091e+07\n",
            "35             context_path.features.layer1.2.relu   256 256 128   256 256 128         0.0      32.00        8,388,608.0       8,388,608.0   33554432.0   33554432.0       0.08%  6.710886e+07\n",
            "36            context_path.features.layer2.0.conv1   256 256 128   128 256 128     32768.0      16.00    2,143,289,344.0   1,073,741,824.0   33685504.0   16777216.0       1.45%  5.046272e+07\n",
            "37              context_path.features.layer2.0.bn1   128 256 128   128 256 128       256.0      16.00       16,777,216.0       8,388,608.0   16778240.0   16777216.0       0.13%  3.355546e+07\n",
            "38            context_path.features.layer2.0.conv2   128 256 128   128 128  64    147456.0       4.00    2,414,870,528.0   1,207,959,552.0   17367040.0    4194304.0       1.20%  2.156134e+07\n",
            "39              context_path.features.layer2.0.bn2   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "40            context_path.features.layer2.0.conv3   128 128  64   512 128  64     65536.0      16.00    1,069,547,520.0     536,870,912.0    4456448.0   16777216.0       0.49%  2.123366e+07\n",
            "41              context_path.features.layer2.0.bn3   512 128  64   512 128  64      1024.0      16.00       16,777,216.0       8,388,608.0   16781312.0   16777216.0       0.12%  3.355853e+07\n",
            "42             context_path.features.layer2.0.relu   512 128  64   512 128  64         0.0      16.00        4,194,304.0       4,194,304.0   16777216.0   16777216.0       0.05%  3.355443e+07\n",
            "43     context_path.features.layer2.0.downsample.0   256 256 128   512 128  64    131072.0      16.00    2,143,289,344.0   1,073,741,824.0   34078720.0   16777216.0       1.69%  5.085594e+07\n",
            "44     context_path.features.layer2.0.downsample.1   512 128  64   512 128  64      1024.0      16.00       16,777,216.0       8,388,608.0   16781312.0   16777216.0       0.14%  3.355853e+07\n",
            "45            context_path.features.layer2.1.conv1   512 128  64   128 128  64     65536.0       4.00    1,072,693,248.0     536,870,912.0   17039360.0    4194304.0       0.90%  2.123366e+07\n",
            "46              context_path.features.layer2.1.bn1   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "47            context_path.features.layer2.1.conv2   128 128  64   128 128  64    147456.0       4.00    2,414,870,528.0   1,207,959,552.0    4784128.0    4194304.0       0.64%  8.978432e+06\n",
            "48              context_path.features.layer2.1.bn2   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "49            context_path.features.layer2.1.conv3   128 128  64   512 128  64     65536.0      16.00    1,069,547,520.0     536,870,912.0    4456448.0   16777216.0       0.46%  2.123366e+07\n",
            "50              context_path.features.layer2.1.bn3   512 128  64   512 128  64      1024.0      16.00       16,777,216.0       8,388,608.0   16781312.0   16777216.0       0.12%  3.355853e+07\n",
            "51             context_path.features.layer2.1.relu   512 128  64   512 128  64         0.0      16.00        4,194,304.0       4,194,304.0   16777216.0   16777216.0       0.07%  3.355443e+07\n",
            "52            context_path.features.layer2.2.conv1   512 128  64   128 128  64     65536.0       4.00    1,072,693,248.0     536,870,912.0   17039360.0    4194304.0       1.08%  2.123366e+07\n",
            "53              context_path.features.layer2.2.bn1   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "54            context_path.features.layer2.2.conv2   128 128  64   128 128  64    147456.0       4.00    2,414,870,528.0   1,207,959,552.0    4784128.0    4194304.0       1.08%  8.978432e+06\n",
            "55              context_path.features.layer2.2.bn2   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "56            context_path.features.layer2.2.conv3   128 128  64   512 128  64     65536.0      16.00    1,069,547,520.0     536,870,912.0    4456448.0   16777216.0       0.51%  2.123366e+07\n",
            "57              context_path.features.layer2.2.bn3   512 128  64   512 128  64      1024.0      16.00       16,777,216.0       8,388,608.0   16781312.0   16777216.0       0.13%  3.355853e+07\n",
            "58             context_path.features.layer2.2.relu   512 128  64   512 128  64         0.0      16.00        4,194,304.0       4,194,304.0   16777216.0   16777216.0       0.05%  3.355443e+07\n",
            "59            context_path.features.layer2.3.conv1   512 128  64   128 128  64     65536.0       4.00    1,072,693,248.0     536,870,912.0   17039360.0    4194304.0       0.92%  2.123366e+07\n",
            "60              context_path.features.layer2.3.bn1   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.07%  8.389632e+06\n",
            "61            context_path.features.layer2.3.conv2   128 128  64   128 128  64    147456.0       4.00    2,414,870,528.0   1,207,959,552.0    4784128.0    4194304.0       1.15%  8.978432e+06\n",
            "62              context_path.features.layer2.3.bn2   128 128  64   128 128  64       256.0       4.00        4,194,304.0       2,097,152.0    4195328.0    4194304.0       0.06%  8.389632e+06\n",
            "63            context_path.features.layer2.3.conv3   128 128  64   512 128  64     65536.0      16.00    1,069,547,520.0     536,870,912.0    4456448.0   16777216.0       0.56%  2.123366e+07\n",
            "64              context_path.features.layer2.3.bn3   512 128  64   512 128  64      1024.0      16.00       16,777,216.0       8,388,608.0   16781312.0   16777216.0       0.13%  3.355853e+07\n",
            "65             context_path.features.layer2.3.relu   512 128  64   512 128  64         0.0      16.00        4,194,304.0       4,194,304.0   16777216.0   16777216.0       0.05%  3.355443e+07\n",
            "66            context_path.features.layer3.0.conv1   512 128  64   256 128  64    131072.0       8.00    2,145,386,496.0   1,073,741,824.0   17301504.0    8388608.0       1.30%  2.569011e+07\n",
            "67              context_path.features.layer3.0.bn1   256 128  64   256 128  64       512.0       8.00        8,388,608.0       4,194,304.0    8390656.0    8388608.0       0.09%  1.677926e+07\n",
            "68            context_path.features.layer3.0.conv2   256 128  64   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0   10747904.0    2097152.0       1.20%  1.284506e+07\n",
            "69              context_path.features.layer3.0.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "70            context_path.features.layer3.0.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "71              context_path.features.layer3.0.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "72             context_path.features.layer3.0.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "73     context_path.features.layer3.0.downsample.0   512 128  64  1024  64  32    524288.0       8.00    2,145,386,496.0   1,073,741,824.0   18874368.0    8388608.0       1.40%  2.726298e+07\n",
            "74     context_path.features.layer3.0.downsample.1  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "75            context_path.features.layer3.1.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.79%  1.153434e+07\n",
            "76              context_path.features.layer3.1.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "77            context_path.features.layer3.1.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       1.00%  6.553600e+06\n",
            "78              context_path.features.layer3.1.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "79            context_path.features.layer3.1.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.40%  1.153434e+07\n",
            "80              context_path.features.layer3.1.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "81             context_path.features.layer3.1.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "82            context_path.features.layer3.2.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.78%  1.153434e+07\n",
            "83              context_path.features.layer3.2.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "84            context_path.features.layer3.2.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.99%  6.553600e+06\n",
            "85              context_path.features.layer3.2.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "86            context_path.features.layer3.2.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "87              context_path.features.layer3.2.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "88             context_path.features.layer3.2.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "89            context_path.features.layer3.3.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.77%  1.153434e+07\n",
            "90              context_path.features.layer3.3.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "91            context_path.features.layer3.3.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       1.02%  6.553600e+06\n",
            "92              context_path.features.layer3.3.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "93            context_path.features.layer3.3.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.44%  1.153434e+07\n",
            "94              context_path.features.layer3.3.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "95             context_path.features.layer3.3.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "96            context_path.features.layer3.4.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.75%  1.153434e+07\n",
            "97              context_path.features.layer3.4.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "98            context_path.features.layer3.4.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.99%  6.553600e+06\n",
            "99              context_path.features.layer3.4.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "100           context_path.features.layer3.4.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.43%  1.153434e+07\n",
            "101             context_path.features.layer3.4.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "102            context_path.features.layer3.4.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "103           context_path.features.layer3.5.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.74%  1.153434e+07\n",
            "104             context_path.features.layer3.5.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "105           context_path.features.layer3.5.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.97%  6.553600e+06\n",
            "106             context_path.features.layer3.5.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "107           context_path.features.layer3.5.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.42%  1.153434e+07\n",
            "108             context_path.features.layer3.5.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "109            context_path.features.layer3.5.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "110           context_path.features.layer3.6.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.80%  1.153434e+07\n",
            "111             context_path.features.layer3.6.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "112           context_path.features.layer3.6.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.95%  6.553600e+06\n",
            "113             context_path.features.layer3.6.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.06%  4.196352e+06\n",
            "114           context_path.features.layer3.6.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.43%  1.153434e+07\n",
            "115             context_path.features.layer3.6.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "116            context_path.features.layer3.6.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "117           context_path.features.layer3.7.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.72%  1.153434e+07\n",
            "118             context_path.features.layer3.7.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "119           context_path.features.layer3.7.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       1.01%  6.553600e+06\n",
            "120             context_path.features.layer3.7.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "121           context_path.features.layer3.7.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.50%  1.153434e+07\n",
            "122             context_path.features.layer3.7.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.10%  1.678541e+07\n",
            "123            context_path.features.layer3.7.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "124           context_path.features.layer3.8.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.73%  1.153434e+07\n",
            "125             context_path.features.layer3.8.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "126           context_path.features.layer3.8.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.93%  6.553600e+06\n",
            "127             context_path.features.layer3.8.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "128           context_path.features.layer3.8.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.43%  1.153434e+07\n",
            "129             context_path.features.layer3.8.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.12%  1.678541e+07\n",
            "130            context_path.features.layer3.8.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.06%  1.677722e+07\n",
            "131           context_path.features.layer3.9.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.69%  1.153434e+07\n",
            "132             context_path.features.layer3.9.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "133           context_path.features.layer3.9.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.92%  6.553600e+06\n",
            "134             context_path.features.layer3.9.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "135           context_path.features.layer3.9.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.43%  1.153434e+07\n",
            "136             context_path.features.layer3.9.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.10%  1.678541e+07\n",
            "137            context_path.features.layer3.9.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.03%  1.677722e+07\n",
            "138          context_path.features.layer3.10.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.67%  1.153434e+07\n",
            "139            context_path.features.layer3.10.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "140          context_path.features.layer3.10.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.91%  6.553600e+06\n",
            "141            context_path.features.layer3.10.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "142          context_path.features.layer3.10.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "143            context_path.features.layer3.10.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "144           context_path.features.layer3.10.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.03%  1.677722e+07\n",
            "145          context_path.features.layer3.11.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.65%  1.153434e+07\n",
            "146            context_path.features.layer3.11.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "147          context_path.features.layer3.11.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.88%  6.553600e+06\n",
            "148            context_path.features.layer3.11.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "149          context_path.features.layer3.11.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "150            context_path.features.layer3.11.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "151           context_path.features.layer3.11.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "152          context_path.features.layer3.12.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.64%  1.153434e+07\n",
            "153            context_path.features.layer3.12.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "154          context_path.features.layer3.12.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.93%  6.553600e+06\n",
            "155            context_path.features.layer3.12.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "156          context_path.features.layer3.12.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.43%  1.153434e+07\n",
            "157            context_path.features.layer3.12.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "158           context_path.features.layer3.12.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "159          context_path.features.layer3.13.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.65%  1.153434e+07\n",
            "160            context_path.features.layer3.13.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "161          context_path.features.layer3.13.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.90%  6.553600e+06\n",
            "162            context_path.features.layer3.13.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "163          context_path.features.layer3.13.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.42%  1.153434e+07\n",
            "164            context_path.features.layer3.13.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "165           context_path.features.layer3.13.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "166          context_path.features.layer3.14.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.66%  1.153434e+07\n",
            "167            context_path.features.layer3.14.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "168          context_path.features.layer3.14.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.90%  6.553600e+06\n",
            "169            context_path.features.layer3.14.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "170          context_path.features.layer3.14.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "171            context_path.features.layer3.14.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "172           context_path.features.layer3.14.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.05%  1.677722e+07\n",
            "173          context_path.features.layer3.15.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.67%  1.153434e+07\n",
            "174            context_path.features.layer3.15.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "175          context_path.features.layer3.15.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.85%  6.553600e+06\n",
            "176            context_path.features.layer3.15.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "177          context_path.features.layer3.15.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.42%  1.153434e+07\n",
            "178            context_path.features.layer3.15.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "179           context_path.features.layer3.15.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "180          context_path.features.layer3.16.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.63%  1.153434e+07\n",
            "181            context_path.features.layer3.16.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "182          context_path.features.layer3.16.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.85%  6.553600e+06\n",
            "183            context_path.features.layer3.16.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "184          context_path.features.layer3.16.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.42%  1.153434e+07\n",
            "185            context_path.features.layer3.16.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "186           context_path.features.layer3.16.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.03%  1.677722e+07\n",
            "187          context_path.features.layer3.17.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.59%  1.153434e+07\n",
            "188            context_path.features.layer3.17.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "189          context_path.features.layer3.17.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.84%  6.553600e+06\n",
            "190            context_path.features.layer3.17.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "191          context_path.features.layer3.17.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "192            context_path.features.layer3.17.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "193           context_path.features.layer3.17.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "194          context_path.features.layer3.18.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.57%  1.153434e+07\n",
            "195            context_path.features.layer3.18.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "196          context_path.features.layer3.18.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.80%  6.553600e+06\n",
            "197            context_path.features.layer3.18.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "198          context_path.features.layer3.18.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "199            context_path.features.layer3.18.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "200           context_path.features.layer3.18.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "201          context_path.features.layer3.19.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.56%  1.153434e+07\n",
            "202            context_path.features.layer3.19.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "203          context_path.features.layer3.19.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.80%  6.553600e+06\n",
            "204            context_path.features.layer3.19.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "205          context_path.features.layer3.19.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.48%  1.153434e+07\n",
            "206            context_path.features.layer3.19.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "207           context_path.features.layer3.19.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.03%  1.677722e+07\n",
            "208          context_path.features.layer3.20.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.56%  1.153434e+07\n",
            "209            context_path.features.layer3.20.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "210          context_path.features.layer3.20.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.78%  6.553600e+06\n",
            "211            context_path.features.layer3.20.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "212          context_path.features.layer3.20.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.42%  1.153434e+07\n",
            "213            context_path.features.layer3.20.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.08%  1.678541e+07\n",
            "214           context_path.features.layer3.20.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "215          context_path.features.layer3.21.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.53%  1.153434e+07\n",
            "216            context_path.features.layer3.21.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "217          context_path.features.layer3.21.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.77%  6.553600e+06\n",
            "218            context_path.features.layer3.21.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "219          context_path.features.layer3.21.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "220            context_path.features.layer3.21.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "221           context_path.features.layer3.21.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "222          context_path.features.layer3.22.conv1  1024  64  32   256  64  32    262144.0       2.00    1,073,217,536.0     536,870,912.0    9437184.0    2097152.0       0.52%  1.153434e+07\n",
            "223            context_path.features.layer3.22.bn1   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "224          context_path.features.layer3.22.conv2   256  64  32   256  64  32    589824.0       2.00    2,415,394,816.0   1,207,959,552.0    4456448.0    2097152.0       0.76%  6.553600e+06\n",
            "225            context_path.features.layer3.22.bn2   256  64  32   256  64  32       512.0       2.00        2,097,152.0       1,048,576.0    2099200.0    2097152.0       0.05%  4.196352e+06\n",
            "226          context_path.features.layer3.22.conv3   256  64  32  1024  64  32    262144.0       8.00    1,071,644,672.0     536,870,912.0    3145728.0    8388608.0       0.41%  1.153434e+07\n",
            "227            context_path.features.layer3.22.bn3  1024  64  32  1024  64  32      2048.0       8.00        8,388,608.0       4,194,304.0    8396800.0    8388608.0       0.09%  1.678541e+07\n",
            "228           context_path.features.layer3.22.relu  1024  64  32  1024  64  32         0.0       8.00        2,097,152.0       2,097,152.0    8388608.0    8388608.0       0.04%  1.677722e+07\n",
            "229           context_path.features.layer4.0.conv1  1024  64  32   512  64  32    524288.0       4.00    2,146,435,072.0   1,073,741,824.0   10485760.0    4194304.0       0.82%  1.468006e+07\n",
            "230             context_path.features.layer4.0.bn1   512  64  32   512  64  32      1024.0       4.00        4,194,304.0       2,097,152.0    4198400.0    4194304.0       0.06%  8.392704e+06\n",
            "231           context_path.features.layer4.0.conv2   512  64  32   512  32  16   2359296.0       1.00    2,415,656,960.0   1,207,959,552.0   13631488.0    1048576.0       0.95%  1.468006e+07\n",
            "232             context_path.features.layer4.0.bn2   512  32  16   512  32  16      1024.0       1.00        1,048,576.0         524,288.0    1052672.0    1048576.0       0.04%  2.101248e+06\n",
            "233           context_path.features.layer4.0.conv3   512  32  16  2048  32  16   1048576.0       4.00    1,072,693,248.0     536,870,912.0    5242880.0    4194304.0       0.42%  9.437184e+06\n",
            "234             context_path.features.layer4.0.bn3  2048  32  16  2048  32  16      4096.0       4.00        4,194,304.0       2,097,152.0    4210688.0    4194304.0       0.06%  8.404992e+06\n",
            "235            context_path.features.layer4.0.relu  2048  32  16  2048  32  16         0.0       4.00        1,048,576.0       1,048,576.0    4194304.0    4194304.0       0.03%  8.388608e+06\n",
            "236    context_path.features.layer4.0.downsample.0  1024  64  32  2048  32  16   2097152.0       4.00    2,146,435,072.0   1,073,741,824.0   16777216.0    4194304.0       0.81%  2.097152e+07\n",
            "237    context_path.features.layer4.0.downsample.1  2048  32  16  2048  32  16      4096.0       4.00        4,194,304.0       2,097,152.0    4210688.0    4194304.0       0.06%  8.404992e+06\n",
            "238           context_path.features.layer4.1.conv1  2048  32  16   512  32  16   1048576.0       1.00    1,073,479,680.0     536,870,912.0    8388608.0    1048576.0       0.47%  9.437184e+06\n",
            "239             context_path.features.layer4.1.bn1   512  32  16   512  32  16      1024.0       1.00        1,048,576.0         524,288.0    1052672.0    1048576.0       0.04%  2.101248e+06\n",
            "240           context_path.features.layer4.1.conv2   512  32  16   512  32  16   2359296.0       1.00    2,415,656,960.0   1,207,959,552.0   10485760.0    1048576.0       0.92%  1.153434e+07\n",
            "241             context_path.features.layer4.1.bn2   512  32  16   512  32  16      1024.0       1.00        1,048,576.0         524,288.0    1052672.0    1048576.0       0.04%  2.101248e+06\n",
            "242           context_path.features.layer4.1.conv3   512  32  16  2048  32  16   1048576.0       4.00    1,072,693,248.0     536,870,912.0    5242880.0    4194304.0       0.43%  9.437184e+06\n",
            "243             context_path.features.layer4.1.bn3  2048  32  16  2048  32  16      4096.0       4.00        4,194,304.0       2,097,152.0    4210688.0    4194304.0       0.06%  8.404992e+06\n",
            "244            context_path.features.layer4.1.relu  2048  32  16  2048  32  16         0.0       4.00        1,048,576.0       1,048,576.0    4194304.0    4194304.0       0.03%  8.388608e+06\n",
            "245           context_path.features.layer4.2.conv1  2048  32  16   512  32  16   1048576.0       1.00    1,073,479,680.0     536,870,912.0    8388608.0    1048576.0       0.50%  9.437184e+06\n",
            "246             context_path.features.layer4.2.bn1   512  32  16   512  32  16      1024.0       1.00        1,048,576.0         524,288.0    1052672.0    1048576.0       0.04%  2.101248e+06\n",
            "247           context_path.features.layer4.2.conv2   512  32  16   512  32  16   2359296.0       1.00    2,415,656,960.0   1,207,959,552.0   10485760.0    1048576.0       0.75%  1.153434e+07\n",
            "248             context_path.features.layer4.2.bn2   512  32  16   512  32  16      1024.0       1.00        1,048,576.0         524,288.0    1052672.0    1048576.0       0.05%  2.101248e+06\n",
            "249           context_path.features.layer4.2.conv3   512  32  16  2048  32  16   1048576.0       4.00    1,072,693,248.0     536,870,912.0    5242880.0    4194304.0       0.43%  9.437184e+06\n",
            "250             context_path.features.layer4.2.bn3  2048  32  16  2048  32  16      4096.0       4.00        4,194,304.0       2,097,152.0    4210688.0    4194304.0       0.06%  8.404992e+06\n",
            "251            context_path.features.layer4.2.relu  2048  32  16  2048  32  16         0.0       4.00        1,048,576.0       1,048,576.0    4194304.0    4194304.0       0.03%  8.388608e+06\n",
            "252                  context_path.features.avgpool     0   0   0     0   0   0         0.0       0.00                0.0               0.0          0.0          0.0       0.00%  0.000000e+00\n",
            "253                       context_path.features.fc     0   0   0     0   0   0         0.0       0.00                0.0               0.0          0.0          0.0       0.00%  0.000000e+00\n",
            "254              attention_refinement_module1.conv  1024   1   1  1024   1   1   1049600.0       0.00        2,097,152.0       1,049,600.0    4202496.0       4096.0       0.44%  4.206592e+06\n",
            "255                attention_refinement_module1.bn  1024   1   1  1024   1   1      2048.0       0.00            4,096.0           2,048.0      12288.0       4096.0       0.04%  1.638400e+04\n",
            "256           attention_refinement_module1.sigmoid  1024   1   1  1024   1   1         0.0       0.00                0.0               0.0          0.0          0.0       0.07%  0.000000e+00\n",
            "257           attention_refinement_module1.avgpool  1024  64  32  1024   1   1         0.0       0.00                0.0               0.0          0.0          0.0       0.16%  0.000000e+00\n",
            "258              attention_refinement_module2.conv  2048   1   1  2048   1   1   4196352.0       0.01        8,388,608.0       4,196,352.0   16793600.0       8192.0       0.40%  1.680179e+07\n",
            "259                attention_refinement_module2.bn  2048   1   1  2048   1   1      4096.0       0.01            8,192.0           4,096.0      24576.0       8192.0       0.04%  3.276800e+04\n",
            "260           attention_refinement_module2.sigmoid  2048   1   1  2048   1   1         0.0       0.01                0.0               0.0          0.0          0.0       0.07%  0.000000e+00\n",
            "261           attention_refinement_module2.avgpool  2048  32  16  2048   1   1         0.0       0.01                0.0               0.0          0.0          0.0       0.11%  0.000000e+00\n",
            "262                                   supervision1     0   0   0     0   0   0         0.0       0.00                0.0               0.0          0.0          0.0       0.00%  0.000000e+00\n",
            "263                                   supervision2     0   0   0     0   0   0         0.0       0.00                0.0               0.0          0.0          0.0       0.00%  0.000000e+00\n",
            "264          feature_fusion_module.convblock.conv1  3328 128  64    19 128  64    569088.0       0.59    9,323,782,144.0   4,661,968,896.0  111328256.0     622592.0       4.53%  1.119508e+08\n",
            "265             feature_fusion_module.convblock.bn    19 128  64    19 128  64        38.0       0.59          622,592.0         311,296.0     622744.0     622592.0       0.05%  1.245336e+06\n",
            "266           feature_fusion_module.convblock.relu    19 128  64    19 128  64         0.0       0.59          155,648.0         155,648.0     622592.0     622592.0       0.03%  1.245184e+06\n",
            "267                    feature_fusion_module.conv1    19   1   1    19   1   1       380.0       0.00              722.0             380.0       1596.0         76.0       0.29%  1.672000e+03\n",
            "268                     feature_fusion_module.relu    19   1   1    19   1   1         0.0       0.00               19.0              19.0         76.0         76.0       0.02%  1.520000e+02\n",
            "269                    feature_fusion_module.conv2    19   1   1    19   1   1       380.0       0.00              722.0             380.0       1596.0         76.0       0.05%  1.672000e+03\n",
            "270                  feature_fusion_module.sigmoid    19   1   1    19   1   1         0.0       0.00                0.0               0.0          0.0          0.0       0.34%  0.000000e+00\n",
            "271                  feature_fusion_module.avgpool    19 128  64    19   1   1         0.0       0.00                0.0               0.0          0.0          0.0       0.07%  0.000000e+00\n",
            "272                                           conv   19 1024 512   19 1024 512       380.0      38.00      378,535,936.0     199,229,440.0   39847408.0   39845888.0       1.76%  7.969330e+07\n",
            "total                                                                           48693786.0    1897.83  183,506,941,367.0  91,913,930,507.0   39847408.0   39845888.0     100.00%  4.287384e+09\n",
            "==============================================================================================================================================================================================\n",
            "Total params: 48,693,786\n",
            "----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\n",
            "Total memory: 1897.83MB\n",
            "Total MAdd: 183.51GMAdd\n",
            "Total Flops: 91.91GFlops\n",
            "Total MemR+W: 3.99GB\n",
            "\n",
            "Model-D without DSC\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "      module name   input shape output shape     params memory(MB)              MAdd             Flops  MemRead(B)  MemWrite(B) duration[%]    MemR+W(B)\n",
            "0           conv1   19 1024 512   64 512 256    19520.0      32.00   5,100,273,664.0   2,558,525,440.0  39923968.0   33554432.0      21.13%   73478400.0\n",
            "1           conv2    64 512 256  128 256 128   131200.0      16.00   8,589,934,592.0   4,299,161,600.0  34079232.0   16777216.0      26.92%   50856448.0\n",
            "2           conv3   128 256 128  256 128  64   524544.0       8.00   8,589,934,592.0   4,297,064,448.0  18875392.0    8388608.0      26.40%   27264000.0\n",
            "3           conv4   256 128  64  512  64  32  2097664.0       4.00   8,589,934,592.0   4,296,015,872.0  16779264.0    4194304.0      23.27%   20973568.0\n",
            "4      classifier   512  64  32    1  32  16     8193.0       0.00       8,388,608.0       4,194,816.0   4227076.0       2048.0       1.93%    4229124.0\n",
            "5      leaky_relu   512  64  32  512  64  32        0.0       4.00               0.0       1,048,576.0   4194304.0    4194304.0       0.34%    8388608.0\n",
            "total                                         2781121.0      64.00  30,878,466,048.0  15,456,010,752.0   4194304.0    4194304.0     100.00%  185190148.0\n",
            "========================================================================================================================================================\n",
            "Total params: 2,781,121\n",
            "--------------------------------------------------------------------------------------------------------------------------------------------------------\n",
            "Total memory: 64.00MB\n",
            "Total MAdd: 30.88GMAdd\n",
            "Total Flops: 15.46GFlops\n",
            "Total MemR+W: 176.61MB\n",
            "\n",
            "Model-D with DSC\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "[MAdd]: LeakyReLU is not supported!\n",
            "      module name   input shape output shape    params memory(MB)             MAdd            Flops  MemRead(B)  MemWrite(B) duration[%]    MemR+W(B)\n",
            "0      depthwise1   19 1024 512   19 512 256     304.0       9.50     77,201,408.0     39,845,888.0  39847104.0    9961472.0      18.51%   49808576.0\n",
            "1      pointwise1    19 512 256   64 512 256    1216.0      32.00    310,378,496.0    159,383,552.0   9966336.0   33554432.0      15.61%   43520768.0\n",
            "2      depthwise2    64 512 256   64 256 128    1024.0       8.00     65,011,712.0     33,554,432.0  33558528.0    8388608.0      14.04%   41947136.0\n",
            "3      pointwise2    64 256 128  128 256 128    8192.0      16.00    532,676,608.0    268,435,456.0   8421376.0   16777216.0      11.45%   25198592.0\n",
            "4      depthwise3   128 256 128  128 128  64    2048.0       4.00     32,505,856.0     16,777,216.0  16785408.0    4194304.0       8.65%   20979712.0\n",
            "5      pointwise3   128 128  64  256 128  64   32768.0       8.00    534,773,760.0    268,435,456.0   4325376.0    8388608.0       9.09%   12713984.0\n",
            "6      depthwise4   256 128  64  256  64  32    4096.0       2.00     16,252,928.0      8,388,608.0   8404992.0    2097152.0       6.19%   10502144.0\n",
            "7      pointwise4   256  64  32  512  64  32  131072.0       4.00    535,822,336.0    268,435,456.0   2621440.0    4194304.0       7.96%    6815744.0\n",
            "8      depthwisec   512  64  32  512  32  16    8192.0       1.00      8,126,464.0      4,194,304.0   4227072.0    1048576.0       5.15%    5275648.0\n",
            "9      pointwisec   512  32  16    1  32  16     512.0       0.00        523,776.0        262,144.0   1050624.0       2048.0       2.21%    1052672.0\n",
            "10     leaky_relu   512  64  32  512  64  32       0.0       4.00              0.0      1,048,576.0   4194304.0    4194304.0       1.13%    8388608.0\n",
            "total                                         189424.0      88.50  2,113,273,344.0  1,068,761,088.0   4194304.0    4194304.0     100.00%  226203584.0\n",
            "=====================================================================================================================================================\n",
            "Total params: 189,424\n",
            "-----------------------------------------------------------------------------------------------------------------------------------------------------\n",
            "Total memory: 88.50MB\n",
            "Total MAdd: 2.11GMAdd\n",
            "Total Flops: 1.07GFlops\n",
            "Total MemR+W: 215.72MB\n",
            "\n"
          ]
        }
      ],
      "source": [
        "#MEASURE STATS FOR STEP 3\n",
        "from torchstat import stat\n",
        "from model.build_BiSeNet import BiSeNet\n",
        "from model.discriminator import FCDiscriminator\n",
        "from model.discriminator_dsc import DiscriminatorDSC\n",
        "\n",
        "print('Model')\n",
        "model=BiSeNet(19, 'resnet101')\n",
        "stat(model, (3,1024,512))\n",
        "print('Model-D without DSC')\n",
        "model_D = FCDiscriminator(19)\n",
        "stat(model_D, (19,1024,512))\n",
        "print('Model-D with DSC')\n",
        "model_D_DSC = DiscriminatorDSC(19)\n",
        "stat(model_D_DSC, (19,1024,512))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "C1gzUBCpFfjC",
        "outputId": "a9333fff-0c4d-4149-d3e0-4f5b22923bad"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch 0, lr 0.025000: 100% 1000/1000 [02:27<00:00,  6.78it/s]\n",
            "loss for train : 3.306235\n",
            "start val!\n",
            "precision per pixel for test: 0.527\n",
            "mIoU for validation: 0.133\n",
            "iter =        0/      50, loss_segmentation = 2.204 loss_adversarial = 0.001 loss_D = 0.724 \n",
            "epoch 1, lr 0.024550: 100% 1000/1000 [02:19<00:00,  7.15it/s]\n",
            "loss for train : 2.244448\n",
            "start val!\n",
            "precision per pixel for test: 0.638\n",
            "mIoU for validation: 0.170\n",
            "iter =        1/      50, loss_segmentation = 1.606 loss_adversarial = 0.001 loss_D = 0.689 \n",
            "epoch 2, lr 0.024098: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 1.775992\n",
            "start val!\n",
            "precision per pixel for test: 0.490\n",
            "mIoU for validation: 0.149\n",
            "iter =        2/      50, loss_segmentation = 1.837 loss_adversarial = 0.001 loss_D = 0.642 \n",
            "epoch 3, lr 0.023646: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 1.623408\n",
            "start val!\n",
            "precision per pixel for test: 0.620\n",
            "mIoU for validation: 0.187\n",
            "iter =        3/      50, loss_segmentation = 1.110 loss_adversarial = 0.001 loss_D = 0.655 \n",
            "epoch 4, lr 0.023193: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 1.439007\n",
            "start val!\n",
            "precision per pixel for test: 0.649\n",
            "mIoU for validation: 0.206\n",
            "iter =        4/      50, loss_segmentation = 1.185 loss_adversarial = 0.001 loss_D = 0.665 \n",
            "epoch 5, lr 0.022738: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 1.274104\n",
            "start val!\n",
            "precision per pixel for test: 0.634\n",
            "mIoU for validation: 0.211\n",
            "iter =        5/      50, loss_segmentation = 0.997 loss_adversarial = 0.001 loss_D = 0.648 \n",
            "epoch 6, lr 0.022283: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 1.227763\n",
            "start val!\n",
            "precision per pixel for test: 0.685\n",
            "mIoU for validation: 0.234\n",
            "iter =        6/      50, loss_segmentation = 0.869 loss_adversarial = 0.001 loss_D = 0.653 \n",
            "epoch 7, lr 0.021827: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 1.049929\n",
            "start val!\n",
            "precision per pixel for test: 0.659\n",
            "mIoU for validation: 0.225\n",
            "iter =        7/      50, loss_segmentation = 0.808 loss_adversarial = 0.001 loss_D = 0.625 \n",
            "epoch 8, lr 0.021369: 100% 1000/1000 [02:20<00:00,  7.10it/s]\n",
            "loss for train : 1.073869\n",
            "start val!\n",
            "precision per pixel for test: 0.629\n",
            "mIoU for validation: 0.205\n",
            "iter =        8/      50, loss_segmentation = 1.207 loss_adversarial = 0.001 loss_D = 0.608 \n",
            "epoch 9, lr 0.020911: 100% 1000/1000 [02:21<00:00,  7.09it/s]\n",
            "loss for train : 0.953992\n",
            "start val!\n",
            "precision per pixel for test: 0.659\n",
            "mIoU for validation: 0.232\n",
            "iter =        9/      50, loss_segmentation = 0.841 loss_adversarial = 0.001 loss_D = 0.626 \n",
            "epoch 10, lr 0.020451: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.867824\n",
            "start val!\n",
            "precision per pixel for test: 0.658\n",
            "mIoU for validation: 0.229\n",
            "iter =       10/      50, loss_segmentation = 0.546 loss_adversarial = 0.001 loss_D = 0.608 \n",
            "epoch 11, lr 0.019991: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.839914\n",
            "start val!\n",
            "precision per pixel for test: 0.646\n",
            "mIoU for validation: 0.235\n",
            "iter =       11/      50, loss_segmentation = 0.935 loss_adversarial = 0.001 loss_D = 0.647 \n",
            "epoch 12, lr 0.019529: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.711707\n",
            "start val!\n",
            "precision per pixel for test: 0.699\n",
            "mIoU for validation: 0.270\n",
            "iter =       12/      50, loss_segmentation = 0.626 loss_adversarial = 0.001 loss_D = 0.609 \n",
            "epoch 13, lr 0.019066: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.655182\n",
            "start val!\n",
            "precision per pixel for test: 0.672\n",
            "mIoU for validation: 0.245\n",
            "iter =       13/      50, loss_segmentation = 0.715 loss_adversarial = 0.001 loss_D = 0.634 \n",
            "epoch 14, lr 0.018601: 100% 1000/1000 [02:20<00:00,  7.10it/s]\n",
            "loss for train : 0.740008\n",
            "start val!\n",
            "precision per pixel for test: 0.666\n",
            "mIoU for validation: 0.245\n",
            "iter =       14/      50, loss_segmentation = 0.695 loss_adversarial = 0.001 loss_D = 0.601 \n",
            "epoch 15, lr 0.018135: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.595584\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.271\n",
            "iter =       15/      50, loss_segmentation = 0.445 loss_adversarial = 0.001 loss_D = 0.562 \n",
            "epoch 16, lr 0.017668: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.545467\n",
            "start val!\n",
            "precision per pixel for test: 0.688\n",
            "mIoU for validation: 0.278\n",
            "iter =       16/      50, loss_segmentation = 0.553 loss_adversarial = 0.001 loss_D = 0.775 \n",
            "epoch 17, lr 0.017200: 100% 1000/1000 [02:21<00:00,  7.09it/s]\n",
            "loss for train : 0.508954\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.276\n",
            "iter =       17/      50, loss_segmentation = 0.559 loss_adversarial = 0.001 loss_D = 0.560 \n",
            "epoch 18, lr 0.016730: 100% 1000/1000 [02:20<00:00,  7.10it/s]\n",
            "loss for train : 0.491993\n",
            "start val!\n",
            "precision per pixel for test: 0.667\n",
            "mIoU for validation: 0.274\n",
            "iter =       18/      50, loss_segmentation = 0.449 loss_adversarial = 0.001 loss_D = 0.539 \n",
            "epoch 19, lr 0.016259: 100% 1000/1000 [02:20<00:00,  7.09it/s]\n",
            "loss for train : 0.468966\n",
            "start val!\n",
            "precision per pixel for test: 0.713\n",
            "mIoU for validation: 0.286\n",
            "iter =       19/      50, loss_segmentation = 0.423 loss_adversarial = 0.001 loss_D = 0.481 \n",
            "epoch 20, lr 0.015786: 100% 1000/1000 [02:20<00:00,  7.10it/s]\n",
            "loss for train : 0.452982\n",
            "start val!\n",
            "precision per pixel for test: 0.702\n",
            "mIoU for validation: 0.283\n",
            "iter =       20/      50, loss_segmentation = 0.415 loss_adversarial = 0.001 loss_D = 0.492 \n",
            "epoch 21, lr 0.015312: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.439892\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.283\n",
            "iter =       21/      50, loss_segmentation = 0.588 loss_adversarial = 0.002 loss_D = 0.544 \n",
            "epoch 22, lr 0.014836: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.429971\n",
            "start val!\n",
            "precision per pixel for test: 0.705\n",
            "mIoU for validation: 0.287\n",
            "iter =       22/      50, loss_segmentation = 0.437 loss_adversarial = 0.001 loss_D = 0.508 \n",
            "epoch 23, lr 0.014358: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.437111\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.283\n",
            "iter =       23/      50, loss_segmentation = 0.509 loss_adversarial = 0.002 loss_D = 0.456 \n",
            "epoch 24, lr 0.013879: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.424641\n",
            "start val!\n",
            "precision per pixel for test: 0.705\n",
            "mIoU for validation: 0.285\n",
            "iter =       24/      50, loss_segmentation = 0.391 loss_adversarial = 0.002 loss_D = 0.580 \n",
            "epoch 25, lr 0.013397: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.408156\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.295\n",
            "iter =       25/      50, loss_segmentation = 0.400 loss_adversarial = 0.002 loss_D = 0.491 \n",
            "epoch 26, lr 0.012914: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.450639\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.276\n",
            "iter =       26/      50, loss_segmentation = 0.483 loss_adversarial = 0.001 loss_D = 0.501 \n",
            "epoch 27, lr 0.012429: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.449130\n",
            "start val!\n",
            "precision per pixel for test: 0.670\n",
            "mIoU for validation: 0.269\n",
            "iter =       27/      50, loss_segmentation = 0.476 loss_adversarial = 0.001 loss_D = 0.521 \n",
            "epoch 28, lr 0.011941: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.407194\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.289\n",
            "iter =       28/      50, loss_segmentation = 0.335 loss_adversarial = 0.001 loss_D = 0.477 \n",
            "epoch 29, lr 0.011452: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.384394\n",
            "start val!\n",
            "precision per pixel for test: 0.686\n",
            "mIoU for validation: 0.279\n",
            "iter =       29/      50, loss_segmentation = 0.399 loss_adversarial = 0.002 loss_D = 0.458 \n",
            "epoch 30, lr 0.010960: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.374058\n",
            "start val!\n",
            "precision per pixel for test: 0.703\n",
            "mIoU for validation: 0.286\n",
            "iter =       30/      50, loss_segmentation = 0.375 loss_adversarial = 0.002 loss_D = 0.482 \n",
            "epoch 31, lr 0.010465: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.368044\n",
            "start val!\n",
            "precision per pixel for test: 0.705\n",
            "mIoU for validation: 0.290\n",
            "iter =       31/      50, loss_segmentation = 0.284 loss_adversarial = 0.001 loss_D = 0.556 \n",
            "epoch 32, lr 0.009968: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.362166\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.294\n",
            "iter =       32/      50, loss_segmentation = 0.364 loss_adversarial = 0.001 loss_D = 0.596 \n",
            "epoch 33, lr 0.009468: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.356213\n",
            "start val!\n",
            "precision per pixel for test: 0.702\n",
            "mIoU for validation: 0.285\n",
            "iter =       33/      50, loss_segmentation = 0.424 loss_adversarial = 0.001 loss_D = 0.594 \n",
            "epoch 34, lr 0.008966: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.352025\n",
            "start val!\n",
            "precision per pixel for test: 0.709\n",
            "mIoU for validation: 0.290\n",
            "iter =       34/      50, loss_segmentation = 0.333 loss_adversarial = 0.002 loss_D = 0.487 \n",
            "epoch 35, lr 0.008460: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.347007\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.289\n",
            "iter =       35/      50, loss_segmentation = 0.324 loss_adversarial = 0.002 loss_D = 0.556 \n",
            "epoch 36, lr 0.007950: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.343463\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.294\n",
            "iter =       36/      50, loss_segmentation = 0.372 loss_adversarial = 0.001 loss_D = 0.560 \n",
            "epoch 37, lr 0.007437: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.341586\n",
            "start val!\n",
            "precision per pixel for test: 0.700\n",
            "mIoU for validation: 0.288\n",
            "iter =       37/      50, loss_segmentation = 0.287 loss_adversarial = 0.002 loss_D = 0.422 \n",
            "epoch 38, lr 0.006920: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.337608\n",
            "start val!\n",
            "precision per pixel for test: 0.706\n",
            "mIoU for validation: 0.296\n",
            "iter =       38/      50, loss_segmentation = 0.370 loss_adversarial = 0.002 loss_D = 0.458 \n",
            "epoch 39, lr 0.006399: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.334520\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.296\n",
            "iter =       39/      50, loss_segmentation = 0.407 loss_adversarial = 0.001 loss_D = 0.526 \n",
            "epoch 40, lr 0.005873: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.332513\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.290\n",
            "iter =       40/      50, loss_segmentation = 0.275 loss_adversarial = 0.001 loss_D = 0.536 \n",
            "epoch 41, lr 0.005342: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.328985\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.295\n",
            "iter =       41/      50, loss_segmentation = 0.308 loss_adversarial = 0.002 loss_D = 0.500 \n",
            "epoch 42, lr 0.004804: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.328265\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.291\n",
            "iter =       42/      50, loss_segmentation = 0.407 loss_adversarial = 0.002 loss_D = 0.487 \n",
            "epoch 43, lr 0.004260: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.326054\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.296\n",
            "iter =       43/      50, loss_segmentation = 0.307 loss_adversarial = 0.001 loss_D = 0.513 \n",
            "epoch 44, lr 0.003709: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.323765\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.296\n",
            "iter =       44/      50, loss_segmentation = 0.297 loss_adversarial = 0.002 loss_D = 0.442 \n",
            "epoch 45, lr 0.003147: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.322429\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.294\n",
            "iter =       45/      50, loss_segmentation = 0.336 loss_adversarial = 0.002 loss_D = 0.456 \n",
            "epoch 46, lr 0.002575: 100% 1000/1000 [02:20<00:00,  7.13it/s]\n",
            "loss for train : 0.320679\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.295\n",
            "iter =       46/      50, loss_segmentation = 0.313 loss_adversarial = 0.001 loss_D = 0.567 \n",
            "epoch 47, lr 0.001987: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.319963\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.296\n",
            "iter =       47/      50, loss_segmentation = 0.287 loss_adversarial = 0.002 loss_D = 0.440 \n",
            "epoch 48, lr 0.001380: 100% 1000/1000 [02:20<00:00,  7.11it/s]\n",
            "loss for train : 0.325332\n",
            "start val!\n",
            "precision per pixel for test: 0.707\n",
            "mIoU for validation: 0.290\n",
            "iter =       48/      50, loss_segmentation = 0.291 loss_adversarial = 0.002 loss_D = 0.494 \n",
            "epoch 49, lr 0.000739: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "loss for train : 0.318583\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.292\n",
            "iter =       49/      50, loss_segmentation = 0.268 loss_adversarial = 0.002 loss_D = 0.541 \n"
          ]
        }
      ],
      "source": [
        "#DOMAIN ADAPTATION, WITHOUT DSC\n",
        "!python trainGTAtoCityscapes.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tPZseiKntD48",
        "outputId": "3cad741b-c809-48e9-d92c-ce770f1880fc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd_da/best_crossentropy_loss.pth ...\n",
            "The best epoch is 47\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:37<00:00,  1.35s/it]\n",
            "IoU for class road: 0.855\n",
            "IoU for class sidewalk: 0.300\n",
            "IoU for class building: 0.760\n",
            "IoU for class wall: 0.219\n",
            "IoU for class fence: 0.069\n",
            "IoU for class pole: 0.176\n",
            "IoU for class light: 0.100\n",
            "IoU for class sign: 0.029\n",
            "IoU for class vegetation: 0.803\n",
            "IoU for class terrain: 0.219\n",
            "IoU for class sky: 0.721\n",
            "IoU for class person: 0.363\n",
            "IoU for class rider: 0.079\n",
            "IoU for class car: 0.661\n",
            "IoU for class truck: 0.147\n",
            "IoU for class bus: 0.042\n",
            "IoU for class train: 0.000\n",
            "IoU for class motocycle: 0.058\n",
            "IoU for class bicycle: 0.032\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.296\n"
          ]
        }
      ],
      "source": [
        "#DOMAIN ADAPTATION, WITHOUT DSC\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Dhs0CfaiCrsJ",
        "outputId": "00fb3a07-587f-4820-dcc3-853f25244680"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch 0, lr 0.025000: 100% 1000/1000 [02:17<00:00,  7.26it/s]\n",
            "loss for train : 3.236787\n",
            "start val!\n",
            "precision per pixel for test: 0.425\n",
            "mIoU for validation: 0.094\n",
            "iter =        0/      50, loss_segmentation = 3.370 loss_adversarial = 0.665 loss_D = 0.691 \n",
            "epoch 1, lr 0.024550: 100% 1000/1000 [02:16<00:00,  7.32it/s]\n",
            "loss for train : 1.966524\n",
            "start val!\n",
            "precision per pixel for test: 0.667\n",
            "mIoU for validation: 0.195\n",
            "iter =        1/      50, loss_segmentation = 1.446 loss_adversarial = 0.715 loss_D = 0.670 \n",
            "epoch 2, lr 0.024098: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 1.757367\n",
            "start val!\n",
            "precision per pixel for test: 0.668\n",
            "mIoU for validation: 0.217\n",
            "iter =        2/      50, loss_segmentation = 1.560 loss_adversarial = 0.736 loss_D = 0.672 \n",
            "epoch 3, lr 0.023646: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 1.444716\n",
            "start val!\n",
            "precision per pixel for test: 0.643\n",
            "mIoU for validation: 0.229\n",
            "iter =        3/      50, loss_segmentation = 1.178 loss_adversarial = 0.717 loss_D = 0.665 \n",
            "epoch 4, lr 0.023193: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 1.250548\n",
            "start val!\n",
            "precision per pixel for test: 0.666\n",
            "mIoU for validation: 0.227\n",
            "iter =        4/      50, loss_segmentation = 1.150 loss_adversarial = 0.857 loss_D = 0.651 \n",
            "epoch 5, lr 0.022738: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 1.073966\n",
            "start val!\n",
            "precision per pixel for test: 0.691\n",
            "mIoU for validation: 0.245\n",
            "iter =        5/      50, loss_segmentation = 1.653 loss_adversarial = 0.772 loss_D = 0.685 \n",
            "epoch 6, lr 0.022283: 100% 1000/1000 [02:17<00:00,  7.27it/s]\n",
            "loss for train : 1.069176\n",
            "start val!\n",
            "precision per pixel for test: 0.592\n",
            "mIoU for validation: 0.201\n",
            "iter =        6/      50, loss_segmentation = 0.991 loss_adversarial = 0.829 loss_D = 0.672 \n",
            "epoch 7, lr 0.021827: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.959928\n",
            "start val!\n",
            "precision per pixel for test: 0.685\n",
            "mIoU for validation: 0.255\n",
            "iter =        7/      50, loss_segmentation = 0.663 loss_adversarial = 0.743 loss_D = 0.674 \n",
            "epoch 8, lr 0.021369: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.802852\n",
            "start val!\n",
            "precision per pixel for test: 0.704\n",
            "mIoU for validation: 0.275\n",
            "iter =        8/      50, loss_segmentation = 0.748 loss_adversarial = 0.774 loss_D = 0.650 \n",
            "epoch 9, lr 0.020911: 100% 1000/1000 [02:17<00:00,  7.28it/s]\n",
            "loss for train : 0.790418\n",
            "start val!\n",
            "precision per pixel for test: 0.529\n",
            "mIoU for validation: 0.201\n",
            "iter =        9/      50, loss_segmentation = 0.839 loss_adversarial = 0.789 loss_D = 0.598 \n",
            "epoch 10, lr 0.020451: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.752693\n",
            "start val!\n",
            "precision per pixel for test: 0.603\n",
            "mIoU for validation: 0.224\n",
            "iter =       10/      50, loss_segmentation = 0.838 loss_adversarial = 0.741 loss_D = 0.696 \n",
            "epoch 11, lr 0.019991: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.625052\n",
            "start val!\n",
            "precision per pixel for test: 0.681\n",
            "mIoU for validation: 0.264\n",
            "iter =       11/      50, loss_segmentation = 0.575 loss_adversarial = 0.956 loss_D = 0.629 \n",
            "epoch 12, lr 0.019529: 100% 1000/1000 [02:16<00:00,  7.32it/s]\n",
            "loss for train : 0.580394\n",
            "start val!\n",
            "precision per pixel for test: 0.683\n",
            "mIoU for validation: 0.268\n",
            "iter =       12/      50, loss_segmentation = 0.523 loss_adversarial = 1.044 loss_D = 0.602 \n",
            "epoch 13, lr 0.019066: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.539433\n",
            "start val!\n",
            "precision per pixel for test: 0.653\n",
            "mIoU for validation: 0.255\n",
            "iter =       13/      50, loss_segmentation = 0.761 loss_adversarial = 0.929 loss_D = 0.592 \n",
            "epoch 14, lr 0.018601: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.506259\n",
            "start val!\n",
            "precision per pixel for test: 0.699\n",
            "mIoU for validation: 0.278\n",
            "iter =       14/      50, loss_segmentation = 0.447 loss_adversarial = 1.052 loss_D = 0.592 \n",
            "epoch 15, lr 0.018135: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.505886\n",
            "start val!\n",
            "precision per pixel for test: 0.644\n",
            "mIoU for validation: 0.248\n",
            "iter =       15/      50, loss_segmentation = 0.403 loss_adversarial = 0.969 loss_D = 0.604 \n",
            "epoch 16, lr 0.017668: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.498106\n",
            "start val!\n",
            "precision per pixel for test: 0.692\n",
            "mIoU for validation: 0.265\n",
            "iter =       16/      50, loss_segmentation = 0.526 loss_adversarial = 0.931 loss_D = 0.621 \n",
            "epoch 17, lr 0.017200: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.476664\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.272\n",
            "iter =       17/      50, loss_segmentation = 0.561 loss_adversarial = 0.952 loss_D = 0.625 \n",
            "epoch 18, lr 0.016730: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.627072\n",
            "start val!\n",
            "precision per pixel for test: 0.666\n",
            "mIoU for validation: 0.267\n",
            "iter =       18/      50, loss_segmentation = 0.464 loss_adversarial = 0.907 loss_D = 0.605 \n",
            "epoch 19, lr 0.016259: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.486188\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.280\n",
            "iter =       19/      50, loss_segmentation = 0.446 loss_adversarial = 1.040 loss_D = 0.572 \n",
            "epoch 20, lr 0.015786: 100% 1000/1000 [02:17<00:00,  7.28it/s]\n",
            "loss for train : 0.459777\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.277\n",
            "iter =       20/      50, loss_segmentation = 0.572 loss_adversarial = 1.053 loss_D = 0.639 \n",
            "epoch 21, lr 0.015312: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.423276\n",
            "start val!\n",
            "precision per pixel for test: 0.695\n",
            "mIoU for validation: 0.278\n",
            "iter =       21/      50, loss_segmentation = 0.423 loss_adversarial = 0.943 loss_D = 0.666 \n",
            "epoch 22, lr 0.014836: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.426648\n",
            "start val!\n",
            "precision per pixel for test: 0.653\n",
            "mIoU for validation: 0.260\n",
            "iter =       22/      50, loss_segmentation = 0.481 loss_adversarial = 1.227 loss_D = 0.578 \n",
            "epoch 23, lr 0.014358: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.416191\n",
            "start val!\n",
            "precision per pixel for test: 0.684\n",
            "mIoU for validation: 0.283\n",
            "iter =       23/      50, loss_segmentation = 0.564 loss_adversarial = 1.068 loss_D = 0.577 \n",
            "epoch 24, lr 0.013879: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.394215\n",
            "start val!\n",
            "precision per pixel for test: 0.691\n",
            "mIoU for validation: 0.280\n",
            "iter =       24/      50, loss_segmentation = 0.400 loss_adversarial = 1.437 loss_D = 0.505 \n",
            "epoch 25, lr 0.013397: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.406350\n",
            "start val!\n",
            "precision per pixel for test: 0.691\n",
            "mIoU for validation: 0.274\n",
            "iter =       25/      50, loss_segmentation = 0.465 loss_adversarial = 1.201 loss_D = 0.536 \n",
            "epoch 26, lr 0.012914: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.397354\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.289\n",
            "iter =       26/      50, loss_segmentation = 0.370 loss_adversarial = 0.862 loss_D = 0.593 \n",
            "epoch 27, lr 0.012429: 100% 1000/1000 [02:17<00:00,  7.28it/s]\n",
            "loss for train : 0.374965\n",
            "start val!\n",
            "precision per pixel for test: 0.703\n",
            "mIoU for validation: 0.290\n",
            "iter =       27/      50, loss_segmentation = 0.422 loss_adversarial = 0.973 loss_D = 0.644 \n",
            "epoch 28, lr 0.011941: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.369633\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.286\n",
            "iter =       28/      50, loss_segmentation = 0.340 loss_adversarial = 1.013 loss_D = 0.604 \n",
            "epoch 29, lr 0.011452: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.360154\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.279\n",
            "iter =       29/      50, loss_segmentation = 0.369 loss_adversarial = 0.881 loss_D = 0.617 \n",
            "epoch 30, lr 0.010960: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.356792\n",
            "start val!\n",
            "precision per pixel for test: 0.689\n",
            "mIoU for validation: 0.280\n",
            "iter =       30/      50, loss_segmentation = 0.278 loss_adversarial = 0.886 loss_D = 0.635 \n",
            "epoch 31, lr 0.010465: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.371340\n",
            "start val!\n",
            "precision per pixel for test: 0.558\n",
            "mIoU for validation: 0.205\n",
            "iter =       31/      50, loss_segmentation = 0.371 loss_adversarial = 1.163 loss_D = 0.615 \n",
            "epoch 32, lr 0.009968: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.421067\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.289\n",
            "iter =       32/      50, loss_segmentation = 0.395 loss_adversarial = 1.173 loss_D = 0.545 \n",
            "epoch 33, lr 0.009468: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.360438\n",
            "start val!\n",
            "precision per pixel for test: 0.695\n",
            "mIoU for validation: 0.282\n",
            "iter =       33/      50, loss_segmentation = 0.411 loss_adversarial = 1.076 loss_D = 0.573 \n",
            "epoch 34, lr 0.008966: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.347910\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.290\n",
            "iter =       34/      50, loss_segmentation = 0.392 loss_adversarial = 1.380 loss_D = 0.529 \n",
            "epoch 35, lr 0.008460: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.342582\n",
            "start val!\n",
            "precision per pixel for test: 0.707\n",
            "mIoU for validation: 0.295\n",
            "iter =       35/      50, loss_segmentation = 0.332 loss_adversarial = 1.029 loss_D = 0.601 \n",
            "epoch 36, lr 0.007950: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.339452\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.298\n",
            "iter =       36/      50, loss_segmentation = 0.217 loss_adversarial = 1.132 loss_D = 0.548 \n",
            "epoch 37, lr 0.007437: 100% 1000/1000 [02:17<00:00,  7.28it/s]\n",
            "loss for train : 0.336065\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.292\n",
            "iter =       37/      50, loss_segmentation = 0.339 loss_adversarial = 1.348 loss_D = 0.534 \n",
            "epoch 38, lr 0.006920: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.338295\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.299\n",
            "iter =       38/      50, loss_segmentation = 0.376 loss_adversarial = 0.893 loss_D = 0.637 \n",
            "epoch 39, lr 0.006399: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.329987\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.294\n",
            "iter =       39/      50, loss_segmentation = 0.263 loss_adversarial = 1.246 loss_D = 0.501 \n",
            "epoch 40, lr 0.005873: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.327340\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.294\n",
            "iter =       40/      50, loss_segmentation = 0.303 loss_adversarial = 1.046 loss_D = 0.591 \n",
            "epoch 41, lr 0.005342: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.324624\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.296\n",
            "iter =       41/      50, loss_segmentation = 0.424 loss_adversarial = 1.279 loss_D = 0.568 \n",
            "epoch 42, lr 0.004804: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.321001\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.295\n",
            "iter =       42/      50, loss_segmentation = 0.296 loss_adversarial = 0.863 loss_D = 0.636 \n",
            "epoch 43, lr 0.004260: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.318987\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.297\n",
            "iter =       43/      50, loss_segmentation = 0.316 loss_adversarial = 1.169 loss_D = 0.579 \n",
            "epoch 44, lr 0.003709: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.317581\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.292\n",
            "iter =       44/      50, loss_segmentation = 0.320 loss_adversarial = 1.029 loss_D = 0.594 \n",
            "epoch 45, lr 0.003147: 100% 1000/1000 [02:17<00:00,  7.30it/s]\n",
            "loss for train : 0.315190\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.296\n",
            "iter =       45/      50, loss_segmentation = 0.250 loss_adversarial = 1.342 loss_D = 0.543 \n",
            "epoch 46, lr 0.002575: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.316377\n",
            "start val!\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.300\n",
            "iter =       46/      50, loss_segmentation = 0.329 loss_adversarial = 1.111 loss_D = 0.577 \n",
            "epoch 47, lr 0.001987: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "loss for train : 0.314144\n",
            "start val!\n",
            "precision per pixel for test: 0.713\n",
            "mIoU for validation: 0.294\n",
            "iter =       47/      50, loss_segmentation = 0.306 loss_adversarial = 1.305 loss_D = 0.574 \n",
            "epoch 48, lr 0.001380: 100% 1000/1000 [02:16<00:00,  7.31it/s]\n",
            "loss for train : 0.314831\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.297\n",
            "iter =       48/      50, loss_segmentation = 0.305 loss_adversarial = 1.207 loss_D = 0.624 \n",
            "epoch 49, lr 0.000739: 100% 1000/1000 [02:16<00:00,  7.30it/s]\n",
            "loss for train : 0.312206\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.296\n",
            "iter =       49/      50, loss_segmentation = 0.408 loss_adversarial = 1.068 loss_D = 0.569 \n"
          ]
        }
      ],
      "source": [
        "#DOMAIN ADAPTATION, WITH DSC\n",
        "!python trainGTAtoCityscapes.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CT-RwtU5CzuL",
        "outputId": "4d4fa218-8f78-4c60-c767-09ca3a4f4444"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd_da_DSC/best_crossentropy_loss.pth ...\n",
            "The best epoch is 46\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:40<00:00,  1.36s/it]\n",
            "IoU for class road: 0.855\n",
            "IoU for class sidewalk: 0.310\n",
            "IoU for class building: 0.757\n",
            "IoU for class wall: 0.214\n",
            "IoU for class fence: 0.085\n",
            "IoU for class pole: 0.178\n",
            "IoU for class light: 0.141\n",
            "IoU for class sign: 0.032\n",
            "IoU for class vegetation: 0.802\n",
            "IoU for class terrain: 0.255\n",
            "IoU for class sky: 0.728\n",
            "IoU for class person: 0.373\n",
            "IoU for class rider: 0.063\n",
            "IoU for class car: 0.661\n",
            "IoU for class truck: 0.148\n",
            "IoU for class bus: 0.054\n",
            "IoU for class train: 0.000\n",
            "IoU for class motocycle: 0.046\n",
            "IoU for class bicycle: 0.000\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.300\n"
          ]
        }
      ],
      "source": [
        "#DOMAIN ADAPTATION, WITH DSC\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fRpu7dwCD6q6"
      },
      "source": [
        "STEP 4"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uE5lPsuD5CVW",
        "outputId": "cdb06487-0021-4349-ca5a-60265873b4ae"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/resnet18-f37072fd.pth\" to /root/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth\n",
            "100% 44.7M/44.7M [00:00<00:00, 278MB/s]\n",
            "Downloading: \"https://download.pytorch.org/models/resnet101-63fe2227.pth\" to /root/.cache/torch/hub/checkpoints/resnet101-63fe2227.pth\n",
            "100% 171M/171M [00:00<00:00, 363MB/s]\n",
            "epoch 0, lr 0.025000: 100% 1000/1000 [05:57<00:00,  2.80it/s]\n",
            "loss for train : 3.096354\n",
            "start val!\n",
            "precision per pixel for test: 0.587\n",
            "mIoU for validation: 0.172\n",
            "iter =        0/      50, loss_segmentation = 2.091 loss_adversarial = 0.682 loss_D = 0.687 \n",
            "epoch 1, lr 0.024550: 100% 1000/1000 [06:20<00:00,  2.63it/s]\n",
            "loss for train : 1.933028\n",
            "start val!\n",
            "precision per pixel for test: 0.641\n",
            "mIoU for validation: 0.198\n",
            "iter =        1/      50, loss_segmentation = 2.252 loss_adversarial = 0.678 loss_D = 0.698 \n",
            "epoch 2, lr 0.024098: 100% 1000/1000 [06:22<00:00,  2.62it/s]\n",
            "loss for train : 1.675695\n",
            "start val!\n",
            "precision per pixel for test: 0.549\n",
            "mIoU for validation: 0.184\n",
            "iter =        2/      50, loss_segmentation = 1.450 loss_adversarial = 0.752 loss_D = 0.660 \n",
            "epoch 3, lr 0.023646: 100% 1000/1000 [06:15<00:00,  2.66it/s]\n",
            "loss for train : 1.396633\n",
            "start val!\n",
            "precision per pixel for test: 0.668\n",
            "mIoU for validation: 0.242\n",
            "iter =        3/      50, loss_segmentation = 1.013 loss_adversarial = 0.710 loss_D = 0.700 \n",
            "epoch 4, lr 0.023193: 100% 1000/1000 [06:10<00:00,  2.70it/s]\n",
            "loss for train : 1.152963\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.258\n",
            "iter =        4/      50, loss_segmentation = 1.264 loss_adversarial = 0.716 loss_D = 0.681 \n",
            "epoch 5, lr 0.022738: 100% 1000/1000 [06:09<00:00,  2.71it/s]\n",
            "loss for train : 1.142640\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.243\n",
            "iter =        5/      50, loss_segmentation = 1.189 loss_adversarial = 0.805 loss_D = 0.664 \n",
            "epoch 6, lr 0.022283: 100% 1000/1000 [06:08<00:00,  2.71it/s]\n",
            "loss for train : 1.026248\n",
            "start val!\n",
            "precision per pixel for test: 0.688\n",
            "mIoU for validation: 0.263\n",
            "iter =        6/      50, loss_segmentation = 1.281 loss_adversarial = 0.809 loss_D = 0.635 \n",
            "epoch 7, lr 0.021827: 100% 1000/1000 [06:05<00:00,  2.73it/s]\n",
            "loss for train : 0.908087\n",
            "start val!\n",
            "precision per pixel for test: 0.670\n",
            "mIoU for validation: 0.245\n",
            "iter =        7/      50, loss_segmentation = 0.762 loss_adversarial = 0.782 loss_D = 0.646 \n",
            "epoch 8, lr 0.021369: 100% 1000/1000 [05:57<00:00,  2.80it/s]\n",
            "loss for train : 0.789607\n",
            "start val!\n",
            "precision per pixel for test: 0.655\n",
            "mIoU for validation: 0.261\n",
            "iter =        8/      50, loss_segmentation = 0.974 loss_adversarial = 0.762 loss_D = 0.680 \n",
            "epoch 9, lr 0.020911: 100% 1000/1000 [06:08<00:00,  2.71it/s]\n",
            "loss for train : 0.877747\n",
            "start val!\n",
            "precision per pixel for test: 0.569\n",
            "mIoU for validation: 0.194\n",
            "iter =        9/      50, loss_segmentation = 0.960 loss_adversarial = 0.886 loss_D = 0.625 \n",
            "epoch 10, lr 0.020451: 100% 1000/1000 [06:11<00:00,  2.69it/s]\n",
            "loss for train : 0.845073\n",
            "start val!\n",
            "precision per pixel for test: 0.691\n",
            "mIoU for validation: 0.250\n",
            "iter =       10/      50, loss_segmentation = 0.769 loss_adversarial = 0.838 loss_D = 0.640 \n",
            "epoch 11, lr 0.019991: 100% 1000/1000 [05:59<00:00,  2.78it/s]\n",
            "loss for train : 0.701586\n",
            "start val!\n",
            "precision per pixel for test: 0.668\n",
            "mIoU for validation: 0.247\n",
            "iter =       11/      50, loss_segmentation = 0.605 loss_adversarial = 0.819 loss_D = 0.652 \n",
            "epoch 12, lr 0.019529: 100% 1000/1000 [05:59<00:00,  2.78it/s]\n",
            "loss for train : 0.666210\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.287\n",
            "iter =       12/      50, loss_segmentation = 0.523 loss_adversarial = 0.795 loss_D = 0.614 \n",
            "epoch 13, lr 0.019066: 100% 1000/1000 [05:55<00:00,  2.81it/s]\n",
            "loss for train : 0.585582\n",
            "start val!\n",
            "precision per pixel for test: 0.657\n",
            "mIoU for validation: 0.262\n",
            "iter =       13/      50, loss_segmentation = 0.544 loss_adversarial = 0.788 loss_D = 0.625 \n",
            "epoch 14, lr 0.018601: 100% 1000/1000 [05:50<00:00,  2.86it/s]\n",
            "loss for train : 0.547112\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.284\n",
            "iter =       14/      50, loss_segmentation = 0.551 loss_adversarial = 0.819 loss_D = 0.639 \n",
            "epoch 15, lr 0.018135: 100% 1000/1000 [05:44<00:00,  2.90it/s]\n",
            "loss for train : 0.502195\n",
            "start val!\n",
            "precision per pixel for test: 0.702\n",
            "mIoU for validation: 0.293\n",
            "iter =       15/      50, loss_segmentation = 0.528 loss_adversarial = 0.950 loss_D = 0.611 \n",
            "epoch 16, lr 0.017668: 100% 1000/1000 [05:48<00:00,  2.87it/s]\n",
            "loss for train : 0.504567\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.297\n",
            "iter =       16/      50, loss_segmentation = 0.460 loss_adversarial = 1.024 loss_D = 0.585 \n",
            "epoch 17, lr 0.017200: 100% 1000/1000 [05:45<00:00,  2.90it/s]\n",
            "loss for train : 0.466216\n",
            "start val!\n",
            "precision per pixel for test: 0.692\n",
            "mIoU for validation: 0.292\n",
            "iter =       17/      50, loss_segmentation = 0.468 loss_adversarial = 1.186 loss_D = 0.596 \n",
            "epoch 18, lr 0.016730: 100% 1000/1000 [05:41<00:00,  2.93it/s]\n",
            "loss for train : 0.449842\n",
            "start val!\n",
            "precision per pixel for test: 0.707\n",
            "mIoU for validation: 0.293\n",
            "iter =       18/      50, loss_segmentation = 0.385 loss_adversarial = 0.966 loss_D = 0.600 \n",
            "epoch 19, lr 0.016259: 100% 1000/1000 [05:36<00:00,  2.97it/s]\n",
            "loss for train : 0.429730\n",
            "start val!\n",
            "precision per pixel for test: 0.703\n",
            "mIoU for validation: 0.295\n",
            "iter =       19/      50, loss_segmentation = 0.530 loss_adversarial = 0.901 loss_D = 0.652 \n",
            "epoch 20, lr 0.015786: 100% 1000/1000 [05:20<00:00,  3.12it/s]\n",
            "loss for train : 0.415553\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.289\n",
            "iter =       20/      50, loss_segmentation = 0.387 loss_adversarial = 0.841 loss_D = 0.671 \n",
            "epoch 21, lr 0.015312: 100% 1000/1000 [05:19<00:00,  3.13it/s]\n",
            "loss for train : 0.404970\n",
            "start val!\n",
            "precision per pixel for test: 0.706\n",
            "mIoU for validation: 0.296\n",
            "iter =       21/      50, loss_segmentation = 0.344 loss_adversarial = 1.106 loss_D = 0.595 \n",
            "epoch 22, lr 0.014836: 100% 1000/1000 [05:17<00:00,  3.15it/s]\n",
            "loss for train : 0.394734\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.299\n",
            "iter =       22/      50, loss_segmentation = 0.302 loss_adversarial = 0.945 loss_D = 0.623 \n",
            "epoch 23, lr 0.014358: 100% 1000/1000 [05:16<00:00,  3.16it/s]\n",
            "loss for train : 0.385545\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.301\n",
            "iter =       23/      50, loss_segmentation = 0.387 loss_adversarial = 0.969 loss_D = 0.571 \n",
            "epoch 24, lr 0.013879: 100% 1000/1000 [05:15<00:00,  3.17it/s]\n",
            "loss for train : 0.379984\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.298\n",
            "iter =       24/      50, loss_segmentation = 0.440 loss_adversarial = 0.885 loss_D = 0.622 \n",
            "epoch 25, lr 0.013397: 100% 1000/1000 [05:12<00:00,  3.20it/s]\n",
            "loss for train : 0.371472\n",
            "start val!\n",
            "precision per pixel for test: 0.713\n",
            "mIoU for validation: 0.295\n",
            "iter =       25/      50, loss_segmentation = 0.407 loss_adversarial = 1.075 loss_D = 0.593 \n",
            "epoch 26, lr 0.012914: 100% 1000/1000 [05:10<00:00,  3.22it/s]\n",
            "loss for train : 0.364023\n",
            "start val!\n",
            "precision per pixel for test: 0.719\n",
            "mIoU for validation: 0.294\n",
            "iter =       26/      50, loss_segmentation = 0.551 loss_adversarial = 0.882 loss_D = 0.657 \n",
            "epoch 27, lr 0.012429: 100% 1000/1000 [05:10<00:00,  3.22it/s]\n",
            "loss for train : 0.358911\n",
            "start val!\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.302\n",
            "iter =       27/      50, loss_segmentation = 0.350 loss_adversarial = 0.935 loss_D = 0.622 \n",
            "epoch 28, lr 0.011941: 100% 1000/1000 [05:09<00:00,  3.23it/s]\n",
            "loss for train : 0.353551\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.298\n",
            "iter =       28/      50, loss_segmentation = 0.433 loss_adversarial = 1.081 loss_D = 0.625 \n",
            "epoch 29, lr 0.011452: 100% 1000/1000 [05:10<00:00,  3.22it/s]\n",
            "loss for train : 0.353635\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.304\n",
            "iter =       29/      50, loss_segmentation = 0.304 loss_adversarial = 0.885 loss_D = 0.677 \n",
            "epoch 30, lr 0.010960: 100% 1000/1000 [05:08<00:00,  3.24it/s]\n",
            "loss for train : 0.347054\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.300\n",
            "iter =       30/      50, loss_segmentation = 0.407 loss_adversarial = 1.141 loss_D = 0.557 \n",
            "epoch 31, lr 0.010465: 100% 1000/1000 [05:07<00:00,  3.25it/s]\n",
            "loss for train : 0.340232\n",
            "start val!\n",
            "precision per pixel for test: 0.719\n",
            "mIoU for validation: 0.297\n",
            "iter =       31/      50, loss_segmentation = 0.267 loss_adversarial = 1.264 loss_D = 0.581 \n",
            "epoch 32, lr 0.009968: 100% 1000/1000 [05:06<00:00,  3.26it/s]\n",
            "loss for train : 0.336606\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.297\n",
            "iter =       32/      50, loss_segmentation = 0.470 loss_adversarial = 1.118 loss_D = 0.556 \n",
            "epoch 33, lr 0.009468: 100% 1000/1000 [05:05<00:00,  3.28it/s]\n",
            "loss for train : 0.331341\n",
            "start val!\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.295\n",
            "iter =       33/      50, loss_segmentation = 0.453 loss_adversarial = 1.261 loss_D = 0.621 \n",
            "epoch 34, lr 0.008966: 100% 1000/1000 [05:04<00:00,  3.28it/s]\n",
            "loss for train : 0.328777\n",
            "start val!\n",
            "precision per pixel for test: 0.720\n",
            "mIoU for validation: 0.301\n",
            "iter =       34/      50, loss_segmentation = 0.290 loss_adversarial = 1.192 loss_D = 0.577 \n",
            "epoch 35, lr 0.008460: 100% 1000/1000 [05:05<00:00,  3.28it/s]\n",
            "loss for train : 0.326413\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.298\n",
            "iter =       35/      50, loss_segmentation = 0.268 loss_adversarial = 0.966 loss_D = 0.594 \n",
            "epoch 36, lr 0.007950: 100% 1000/1000 [05:03<00:00,  3.30it/s]\n",
            "loss for train : 0.323648\n",
            "start val!\n",
            "precision per pixel for test: 0.724\n",
            "mIoU for validation: 0.301\n",
            "iter =       36/      50, loss_segmentation = 0.329 loss_adversarial = 0.892 loss_D = 0.680 \n",
            "epoch 37, lr 0.007437: 100% 1000/1000 [05:02<00:00,  3.30it/s]\n",
            "loss for train : 0.319705\n",
            "start val!\n",
            "precision per pixel for test: 0.725\n",
            "mIoU for validation: 0.303\n",
            "iter =       37/      50, loss_segmentation = 0.301 loss_adversarial = 1.084 loss_D = 0.660 \n",
            "epoch 38, lr 0.006920: 100% 1000/1000 [05:01<00:00,  3.32it/s]\n",
            "loss for train : 0.316674\n",
            "start val!\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.298\n",
            "iter =       38/      50, loss_segmentation = 0.257 loss_adversarial = 1.164 loss_D = 0.573 \n",
            "epoch 39, lr 0.006399: 100% 1000/1000 [05:01<00:00,  3.32it/s]\n",
            "loss for train : 0.314712\n",
            "start val!\n",
            "precision per pixel for test: 0.721\n",
            "mIoU for validation: 0.297\n",
            "iter =       39/      50, loss_segmentation = 0.366 loss_adversarial = 0.951 loss_D = 0.622 \n",
            "epoch 40, lr 0.005873: 100% 1000/1000 [05:00<00:00,  3.33it/s]\n",
            "loss for train : 0.312527\n",
            "start val!\n",
            "precision per pixel for test: 0.723\n",
            "mIoU for validation: 0.301\n",
            "iter =       40/      50, loss_segmentation = 0.301 loss_adversarial = 1.206 loss_D = 0.556 \n",
            "epoch 41, lr 0.005342: 100% 1000/1000 [05:01<00:00,  3.32it/s]\n",
            "loss for train : 0.313016\n",
            "start val!\n",
            "precision per pixel for test: 0.724\n",
            "mIoU for validation: 0.298\n",
            "iter =       41/      50, loss_segmentation = 0.276 loss_adversarial = 0.962 loss_D = 0.599 \n",
            "epoch 42, lr 0.004804: 100% 1000/1000 [05:00<00:00,  3.32it/s]\n",
            "loss for train : 0.309305\n",
            "start val!\n",
            "precision per pixel for test: 0.720\n",
            "mIoU for validation: 0.298\n",
            "iter =       42/      50, loss_segmentation = 0.288 loss_adversarial = 1.136 loss_D = 0.608 \n",
            "epoch 43, lr 0.004260: 100% 1000/1000 [04:59<00:00,  3.34it/s]\n",
            "loss for train : 0.307882\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.297\n",
            "iter =       43/      50, loss_segmentation = 0.331 loss_adversarial = 1.191 loss_D = 0.595 \n",
            "epoch 44, lr 0.003709: 100% 1000/1000 [05:00<00:00,  3.33it/s]\n",
            "loss for train : 0.307113\n",
            "start val!\n",
            "precision per pixel for test: 0.719\n",
            "mIoU for validation: 0.299\n",
            "iter =       44/      50, loss_segmentation = 0.417 loss_adversarial = 1.139 loss_D = 0.567 \n",
            "epoch 45, lr 0.003147: 100% 1000/1000 [05:00<00:00,  3.32it/s]\n",
            "loss for train : 0.305787\n",
            "start val!\n",
            "precision per pixel for test: 0.722\n",
            "mIoU for validation: 0.300\n",
            "iter =       45/      50, loss_segmentation = 0.296 loss_adversarial = 1.077 loss_D = 0.548 \n",
            "epoch 46, lr 0.002575: 100% 1000/1000 [04:59<00:00,  3.34it/s]\n",
            "loss for train : 0.304511\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.296\n",
            "iter =       46/      50, loss_segmentation = 0.406 loss_adversarial = 1.199 loss_D = 0.566 \n",
            "epoch 47, lr 0.001987: 100% 1000/1000 [04:59<00:00,  3.34it/s]\n",
            "loss for train : 0.302896\n",
            "start val!\n",
            "precision per pixel for test: 0.720\n",
            "mIoU for validation: 0.297\n",
            "iter =       47/      50, loss_segmentation = 0.339 loss_adversarial = 0.983 loss_D = 0.668 \n",
            "epoch 48, lr 0.001380: 100% 1000/1000 [04:59<00:00,  3.34it/s]\n",
            "loss for train : 0.301925\n",
            "start val!\n",
            "precision per pixel for test: 0.720\n",
            "mIoU for validation: 0.297\n",
            "iter =       48/      50, loss_segmentation = 0.257 loss_adversarial = 1.072 loss_D = 0.637 \n",
            "epoch 49, lr 0.000739: 100% 1000/1000 [04:59<00:00,  3.34it/s]\n",
            "loss for train : 0.301984\n",
            "start val!\n",
            "precision per pixel for test: 0.727\n",
            "mIoU for validation: 0.303\n",
            "iter =       49/      50, loss_segmentation = 0.310 loss_adversarial = 0.849 loss_D = 0.648 \n"
          ]
        }
      ],
      "source": [
        "#a) FDA\n",
        "!python trainGTAtoCityscapes.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wK9aMr7M5Ho9",
        "outputId": "d7269bb6-284c-46c8-9147-eebeaab24af8"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd_da_DSC_FDA/best_crossentropy_loss.pth ...\n",
            "The best epoch is 29\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:19<00:00,  1.28s/it]\n",
            "IoU for class road: 0.832\n",
            "IoU for class sidewalk: 0.347\n",
            "IoU for class building: 0.774\n",
            "IoU for class wall: 0.255\n",
            "IoU for class fence: 0.088\n",
            "IoU for class pole: 0.183\n",
            "IoU for class light: 0.133\n",
            "IoU for class sign: 0.073\n",
            "IoU for class vegetation: 0.797\n",
            "IoU for class terrain: 0.307\n",
            "IoU for class sky: 0.726\n",
            "IoU for class person: 0.348\n",
            "IoU for class rider: 0.003\n",
            "IoU for class car: 0.726\n",
            "IoU for class truck: 0.147\n",
            "IoU for class bus: 0.017\n",
            "IoU for class train: 0.000\n",
            "IoU for class motocycle: 0.022\n",
            "IoU for class bicycle: 0.000\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.304\n"
          ]
        }
      ],
      "source": [
        "#a) FDA\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "63BrB1yGD83M",
        "outputId": "e164bf02-e67f-429d-a326-0bef6e09d4dc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch 0, lr 0.025000: 100% 1000/1000 [03:19<00:00,  5.01it/s]\n",
            "loss for train : 2.982983\n",
            "start val!\n",
            "precision per pixel for test: 0.546\n",
            "mIoU for validation: 0.135\n",
            "iter =        0/      50, loss_segmentation = 1.870 loss_adversarial = 0.693 loss_D = 0.692 \n",
            "epoch 1, lr 0.024550: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 1.952191\n",
            "start val!\n",
            "precision per pixel for test: 0.679\n",
            "mIoU for validation: 0.224\n",
            "iter =        1/      50, loss_segmentation = 1.212 loss_adversarial = 0.697 loss_D = 0.683 \n",
            "epoch 2, lr 0.024098: 100% 1000/1000 [03:21<00:00,  4.96it/s]\n",
            "loss for train : 1.559652\n",
            "start val!\n",
            "precision per pixel for test: 0.659\n",
            "mIoU for validation: 0.231\n",
            "iter =        2/      50, loss_segmentation = 1.774 loss_adversarial = 0.756 loss_D = 0.649 \n",
            "epoch 3, lr 0.023646: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 1.357023\n",
            "start val!\n",
            "precision per pixel for test: 0.578\n",
            "mIoU for validation: 0.196\n",
            "iter =        3/      50, loss_segmentation = 1.106 loss_adversarial = 0.742 loss_D = 0.687 \n",
            "epoch 4, lr 0.023193: 100% 1000/1000 [03:20<00:00,  4.98it/s]\n",
            "loss for train : 1.207217\n",
            "start val!\n",
            "precision per pixel for test: 0.638\n",
            "mIoU for validation: 0.239\n",
            "iter =        4/      50, loss_segmentation = 1.406 loss_adversarial = 0.767 loss_D = 0.688 \n",
            "epoch 5, lr 0.022738: 100% 1000/1000 [03:21<00:00,  4.95it/s]\n",
            "loss for train : 1.171668\n",
            "start val!\n",
            "precision per pixel for test: 0.346\n",
            "mIoU for validation: 0.111\n",
            "iter =        5/      50, loss_segmentation = 0.708 loss_adversarial = 0.735 loss_D = 0.696 \n",
            "epoch 6, lr 0.022283: 100% 1000/1000 [03:20<00:00,  4.99it/s]\n",
            "loss for train : 0.994310\n",
            "start val!\n",
            "precision per pixel for test: 0.627\n",
            "mIoU for validation: 0.209\n",
            "iter =        6/      50, loss_segmentation = 0.860 loss_adversarial = 0.734 loss_D = 0.652 \n",
            "epoch 7, lr 0.021827: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 0.884817\n",
            "start val!\n",
            "precision per pixel for test: 0.638\n",
            "mIoU for validation: 0.249\n",
            "iter =        7/      50, loss_segmentation = 0.959 loss_adversarial = 0.806 loss_D = 0.624 \n",
            "epoch 8, lr 0.021369: 100% 1000/1000 [03:25<00:00,  4.87it/s]\n",
            "loss for train : 0.824794\n",
            "start val!\n",
            "precision per pixel for test: 0.697\n",
            "mIoU for validation: 0.271\n",
            "iter =        8/      50, loss_segmentation = 0.628 loss_adversarial = 0.868 loss_D = 0.649 \n",
            "epoch 9, lr 0.020911: 100% 1000/1000 [03:23<00:00,  4.92it/s]\n",
            "loss for train : 0.687063\n",
            "start val!\n",
            "precision per pixel for test: 0.706\n",
            "mIoU for validation: 0.290\n",
            "iter =        9/      50, loss_segmentation = 0.847 loss_adversarial = 0.882 loss_D = 0.619 \n",
            "epoch 10, lr 0.020451: 100% 1000/1000 [03:21<00:00,  4.95it/s]\n",
            "loss for train : 0.631471\n",
            "start val!\n",
            "precision per pixel for test: 0.631\n",
            "mIoU for validation: 0.274\n",
            "iter =       10/      50, loss_segmentation = 0.789 loss_adversarial = 1.071 loss_D = 0.586 \n",
            "epoch 11, lr 0.019991: 100% 1000/1000 [03:20<00:00,  4.99it/s]\n",
            "loss for train : 0.578456\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.304\n",
            "iter =       11/      50, loss_segmentation = 0.621 loss_adversarial = 0.815 loss_D = 0.611 \n",
            "epoch 12, lr 0.019529: 100% 1000/1000 [03:22<00:00,  4.95it/s]\n",
            "loss for train : 0.613467\n",
            "start val!\n",
            "precision per pixel for test: 0.634\n",
            "mIoU for validation: 0.253\n",
            "iter =       12/      50, loss_segmentation = 0.746 loss_adversarial = 0.808 loss_D = 0.637 \n",
            "epoch 13, lr 0.019066: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 0.589026\n",
            "start val!\n",
            "precision per pixel for test: 0.690\n",
            "mIoU for validation: 0.276\n",
            "iter =       13/      50, loss_segmentation = 1.028 loss_adversarial = 0.814 loss_D = 0.689 \n",
            "epoch 14, lr 0.018601: 100% 1000/1000 [03:23<00:00,  4.92it/s]\n",
            "loss for train : 0.544380\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.292\n",
            "iter =       14/      50, loss_segmentation = 0.431 loss_adversarial = 0.814 loss_D = 0.673 \n",
            "epoch 15, lr 0.018135: 100% 1000/1000 [03:21<00:00,  4.95it/s]\n",
            "loss for train : 0.530004\n",
            "start val!\n",
            "precision per pixel for test: 0.680\n",
            "mIoU for validation: 0.282\n",
            "iter =       15/      50, loss_segmentation = 0.503 loss_adversarial = 0.878 loss_D = 0.575 \n",
            "epoch 16, lr 0.017668: 100% 1000/1000 [03:20<00:00,  4.98it/s]\n",
            "loss for train : 0.628553\n",
            "start val!\n",
            "precision per pixel for test: 0.651\n",
            "mIoU for validation: 0.279\n",
            "iter =       16/      50, loss_segmentation = 0.560 loss_adversarial = 0.932 loss_D = 0.581 \n",
            "epoch 17, lr 0.017200: 100% 1000/1000 [03:23<00:00,  4.92it/s]\n",
            "loss for train : 0.508230\n",
            "start val!\n",
            "precision per pixel for test: 0.704\n",
            "mIoU for validation: 0.294\n",
            "iter =       17/      50, loss_segmentation = 0.462 loss_adversarial = 0.905 loss_D = 0.580 \n",
            "epoch 18, lr 0.016730: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.470048\n",
            "start val!\n",
            "precision per pixel for test: 0.685\n",
            "mIoU for validation: 0.305\n",
            "iter =       18/      50, loss_segmentation = 0.485 loss_adversarial = 0.966 loss_D = 0.619 \n",
            "epoch 19, lr 0.016259: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.436373\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.310\n",
            "iter =       19/      50, loss_segmentation = 0.434 loss_adversarial = 0.932 loss_D = 0.687 \n",
            "epoch 20, lr 0.015786: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.421338\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.312\n",
            "iter =       20/      50, loss_segmentation = 0.349 loss_adversarial = 0.850 loss_D = 0.656 \n",
            "epoch 21, lr 0.015312: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.410972\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.306\n",
            "iter =       21/      50, loss_segmentation = 0.373 loss_adversarial = 0.953 loss_D = 0.583 \n",
            "epoch 22, lr 0.014836: 100% 1000/1000 [03:19<00:00,  5.02it/s]\n",
            "loss for train : 0.399066\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.311\n",
            "iter =       22/      50, loss_segmentation = 0.446 loss_adversarial = 0.986 loss_D = 0.617 \n",
            "epoch 23, lr 0.014358: 100% 1000/1000 [03:18<00:00,  5.04it/s]\n",
            "loss for train : 0.390657\n",
            "start val!\n",
            "precision per pixel for test: 0.719\n",
            "mIoU for validation: 0.313\n",
            "iter =       23/      50, loss_segmentation = 0.556 loss_adversarial = 1.046 loss_D = 0.614 \n",
            "epoch 24, lr 0.013879: 100% 1000/1000 [03:19<00:00,  5.01it/s]\n",
            "loss for train : 0.415571\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.308\n",
            "iter =       24/      50, loss_segmentation = 0.555 loss_adversarial = 1.042 loss_D = 0.594 \n",
            "epoch 25, lr 0.013397: 100% 1000/1000 [03:20<00:00,  5.00it/s]\n",
            "loss for train : 0.385926\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.315\n",
            "iter =       25/      50, loss_segmentation = 0.437 loss_adversarial = 1.018 loss_D = 0.635 \n",
            "epoch 26, lr 0.012914: 100% 1000/1000 [03:19<00:00,  5.02it/s]\n",
            "loss for train : 0.375981\n",
            "start val!\n",
            "precision per pixel for test: 0.709\n",
            "mIoU for validation: 0.307\n",
            "iter =       26/      50, loss_segmentation = 0.401 loss_adversarial = 1.118 loss_D = 0.575 \n",
            "epoch 27, lr 0.012429: 100% 1000/1000 [03:19<00:00,  5.02it/s]\n",
            "loss for train : 0.365499\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.311\n",
            "iter =       27/      50, loss_segmentation = 0.258 loss_adversarial = 1.502 loss_D = 0.513 \n",
            "epoch 28, lr 0.011941: 100% 1000/1000 [03:28<00:00,  4.79it/s]\n",
            "loss for train : 0.358472\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.312\n",
            "iter =       28/      50, loss_segmentation = 0.369 loss_adversarial = 1.112 loss_D = 0.613 \n",
            "epoch 29, lr 0.011452: 100% 1000/1000 [03:20<00:00,  4.98it/s]\n",
            "loss for train : 0.354155\n",
            "start val!\n",
            "precision per pixel for test: 0.707\n",
            "mIoU for validation: 0.307\n",
            "iter =       29/      50, loss_segmentation = 0.289 loss_adversarial = 0.979 loss_D = 0.567 \n",
            "epoch 30, lr 0.010960: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.347575\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.315\n",
            "iter =       30/      50, loss_segmentation = 0.277 loss_adversarial = 1.011 loss_D = 0.561 \n",
            "epoch 31, lr 0.010465: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.343873\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.309\n",
            "iter =       31/      50, loss_segmentation = 0.299 loss_adversarial = 1.126 loss_D = 0.571 \n",
            "epoch 32, lr 0.009968: 100% 1000/1000 [03:21<00:00,  4.95it/s]\n",
            "loss for train : 0.340859\n",
            "start val!\n",
            "precision per pixel for test: 0.701\n",
            "mIoU for validation: 0.300\n",
            "iter =       32/      50, loss_segmentation = 0.425 loss_adversarial = 1.161 loss_D = 0.609 \n",
            "epoch 33, lr 0.009468: 100% 1000/1000 [03:20<00:00,  4.99it/s]\n",
            "loss for train : 0.337317\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.309\n",
            "iter =       33/      50, loss_segmentation = 0.450 loss_adversarial = 0.892 loss_D = 0.662 \n",
            "epoch 34, lr 0.008966: 100% 1000/1000 [03:21<00:00,  4.95it/s]\n",
            "loss for train : 0.332317\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.305\n",
            "iter =       34/      50, loss_segmentation = 0.333 loss_adversarial = 1.166 loss_D = 0.545 \n",
            "epoch 35, lr 0.008460: 100% 1000/1000 [03:21<00:00,  4.97it/s]\n",
            "loss for train : 0.329332\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.305\n",
            "iter =       35/      50, loss_segmentation = 0.383 loss_adversarial = 1.056 loss_D = 0.613 \n",
            "epoch 36, lr 0.007950: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 0.325091\n",
            "start val!\n",
            "precision per pixel for test: 0.719\n",
            "mIoU for validation: 0.311\n",
            "iter =       36/      50, loss_segmentation = 0.313 loss_adversarial = 1.058 loss_D = 0.601 \n",
            "epoch 37, lr 0.007437: 100% 1000/1000 [03:22<00:00,  4.95it/s]\n",
            "loss for train : 0.323411\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.303\n",
            "iter =       37/      50, loss_segmentation = 0.331 loss_adversarial = 0.893 loss_D = 0.705 \n",
            "epoch 38, lr 0.006920: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 0.321522\n",
            "start val!\n",
            "precision per pixel for test: 0.713\n",
            "mIoU for validation: 0.307\n",
            "iter =       38/      50, loss_segmentation = 0.287 loss_adversarial = 0.938 loss_D = 0.597 \n",
            "epoch 39, lr 0.006399: 100% 1000/1000 [03:22<00:00,  4.94it/s]\n",
            "loss for train : 0.318998\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.307\n",
            "iter =       39/      50, loss_segmentation = 0.384 loss_adversarial = 0.936 loss_D = 0.644 \n",
            "epoch 40, lr 0.005873: 100% 1000/1000 [03:22<00:00,  4.93it/s]\n",
            "loss for train : 0.318006\n",
            "start val!\n",
            "precision per pixel for test: 0.710\n",
            "mIoU for validation: 0.308\n",
            "iter =       40/      50, loss_segmentation = 0.291 loss_adversarial = 1.021 loss_D = 0.575 \n",
            "epoch 41, lr 0.005342: 100% 1000/1000 [03:23<00:00,  4.92it/s]\n",
            "loss for train : 0.314972\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.311\n",
            "iter =       41/      50, loss_segmentation = 0.352 loss_adversarial = 1.209 loss_D = 0.563 \n",
            "epoch 42, lr 0.004804: 100% 1000/1000 [03:21<00:00,  4.96it/s]\n",
            "loss for train : 0.312728\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.305\n",
            "iter =       42/      50, loss_segmentation = 0.246 loss_adversarial = 1.175 loss_D = 0.583 \n",
            "epoch 43, lr 0.004260: 100% 1000/1000 [03:23<00:00,  4.92it/s]\n",
            "loss for train : 0.310950\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.305\n",
            "iter =       43/      50, loss_segmentation = 0.294 loss_adversarial = 0.966 loss_D = 0.585 \n",
            "epoch 44, lr 0.003709: 100% 1000/1000 [03:22<00:00,  4.95it/s]\n",
            "loss for train : 0.309952\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.309\n",
            "iter =       44/      50, loss_segmentation = 0.239 loss_adversarial = 1.119 loss_D = 0.598 \n",
            "epoch 45, lr 0.003147: 100% 1000/1000 [03:20<00:00,  4.98it/s]\n",
            "loss for train : 0.309454\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.308\n",
            "iter =       45/      50, loss_segmentation = 0.277 loss_adversarial = 1.204 loss_D = 0.558 \n",
            "epoch 46, lr 0.002575: 100% 1000/1000 [03:32<00:00,  4.71it/s]\n",
            "loss for train : 0.307604\n",
            "start val!\n",
            "precision per pixel for test: 0.716\n",
            "mIoU for validation: 0.304\n",
            "iter =       46/      50, loss_segmentation = 0.396 loss_adversarial = 1.154 loss_D = 0.564 \n",
            "epoch 47, lr 0.001987: 100% 1000/1000 [03:35<00:00,  4.64it/s]\n",
            "loss for train : 0.306961\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.307\n",
            "iter =       47/      50, loss_segmentation = 0.266 loss_adversarial = 1.292 loss_D = 0.574 \n",
            "epoch 48, lr 0.001380: 100% 1000/1000 [03:28<00:00,  4.80it/s]\n",
            "loss for train : 0.305290\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.308\n",
            "iter =       48/      50, loss_segmentation = 0.341 loss_adversarial = 1.110 loss_D = 0.612 \n",
            "epoch 49, lr 0.000739: 100% 1000/1000 [03:28<00:00,  4.79it/s]\n",
            "loss for train : 0.304741\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.306\n",
            "iter =       49/      50, loss_segmentation = 0.291 loss_adversarial = 1.080 loss_D = 0.583 \n"
          ]
        }
      ],
      "source": [
        "#b) LAB\n",
        "!python trainGTAtoCityscapes.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bH9PTCRC4d8v",
        "outputId": "838c971e-6af8-4ed3-bee6-dee326e77806"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd_da_DSC_LAB/best_crossentropy_loss.pth ...\n",
            "The best epoch is 30\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [06:00<00:00,  1.44s/it]\n",
            "IoU for class road: 0.827\n",
            "IoU for class sidewalk: 0.295\n",
            "IoU for class building: 0.782\n",
            "IoU for class wall: 0.273\n",
            "IoU for class fence: 0.094\n",
            "IoU for class pole: 0.192\n",
            "IoU for class light: 0.140\n",
            "IoU for class sign: 0.066\n",
            "IoU for class vegetation: 0.811\n",
            "IoU for class terrain: 0.308\n",
            "IoU for class sky: 0.760\n",
            "IoU for class person: 0.347\n",
            "IoU for class rider: 0.088\n",
            "IoU for class car: 0.724\n",
            "IoU for class truck: 0.159\n",
            "IoU for class bus: 0.063\n",
            "IoU for class train: 0.000\n",
            "IoU for class motocycle: 0.056\n",
            "IoU for class bicycle: 0.003\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.315\n"
          ]
        }
      ],
      "source": [
        "#b) LAB\n",
        "!python eval.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "phfPx8_PEBBk",
        "outputId": "c019daf5-bccd-407a-dc78-078111f90bbb"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "epoch 0, lr 0.025000: 100% 1000/1000 [02:19<00:00,  7.15it/s]\n",
            "average loss for train : 3.053908\n",
            "start val!\n",
            "precision per pixel for test: 0.561\n",
            "mIoU for validation: 0.148\n",
            "iter =        0/      75, loss_segmentation = 1.973 loss_D_trg_fake = 0.676 loss_D = 0.684 loss_seg_trg = 0.000\n",
            "epoch 1, lr 0.024700: 100% 1000/1000 [02:17<00:00,  7.29it/s]\n",
            "average loss for train : 1.842083\n",
            "start val!\n",
            "precision per pixel for test: 0.584\n",
            "mIoU for validation: 0.160\n",
            "iter =        1/      75, loss_segmentation = 1.374 loss_D_trg_fake = 0.744 loss_D = 0.665 loss_seg_trg = 0.000\n",
            "epoch 2, lr 0.024399: 100% 1000/1000 [02:17<00:00,  7.28it/s]\n",
            "average loss for train : 1.522063\n",
            "start val!\n",
            "precision per pixel for test: 0.588\n",
            "mIoU for validation: 0.181\n",
            "iter =        2/      75, loss_segmentation = 2.227 loss_D_trg_fake = 0.736 loss_D = 0.703 loss_seg_trg = 0.000\n",
            "epoch 3, lr 0.024098: 100% 1000/1000 [02:20<00:00,  7.12it/s]\n",
            "average loss for train : 1.353147\n",
            "start val!\n",
            "precision per pixel for test: 0.623\n",
            "mIoU for validation: 0.198\n",
            "iter =        3/      75, loss_segmentation = 0.963 loss_D_trg_fake = 0.771 loss_D = 0.646 loss_seg_trg = 0.000\n",
            "epoch 4, lr 0.023797: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 1.305598\n",
            "start val!\n",
            "precision per pixel for test: 0.642\n",
            "mIoU for validation: 0.203\n",
            "iter =        4/      75, loss_segmentation = 1.260 loss_D_trg_fake = 0.785 loss_D = 0.669 loss_seg_trg = 0.000\n",
            "epoch 5, lr 0.023495: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 1.244784\n",
            "start val!\n",
            "precision per pixel for test: 0.665\n",
            "mIoU for validation: 0.219\n",
            "iter =        5/      75, loss_segmentation = 1.630 loss_D_trg_fake = 0.812 loss_D = 0.624 loss_seg_trg = 0.000\n",
            "epoch 6, lr 0.023193: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 1.040776\n",
            "start val!\n",
            "precision per pixel for test: 0.596\n",
            "mIoU for validation: 0.202\n",
            "iter =        6/      75, loss_segmentation = 0.735 loss_D_trg_fake = 0.748 loss_D = 0.675 loss_seg_trg = 0.000\n",
            "epoch 7, lr 0.022890: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.926550\n",
            "start val!\n",
            "precision per pixel for test: 0.643\n",
            "mIoU for validation: 0.234\n",
            "iter =        7/      75, loss_segmentation = 1.187 loss_D_trg_fake = 0.792 loss_D = 0.643 loss_seg_trg = 0.000\n",
            "epoch 8, lr 0.022587: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.832732\n",
            "start val!\n",
            "precision per pixel for test: 0.615\n",
            "mIoU for validation: 0.221\n",
            "iter =        8/      75, loss_segmentation = 0.791 loss_D_trg_fake = 0.859 loss_D = 0.615 loss_seg_trg = 0.000\n",
            "epoch 9, lr 0.022283: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.767038\n",
            "start val!\n",
            "precision per pixel for test: 0.611\n",
            "mIoU for validation: 0.236\n",
            "iter =        9/      75, loss_segmentation = 0.705 loss_D_trg_fake = 0.914 loss_D = 0.672 loss_seg_trg = 0.000\n",
            "epoch 10, lr 0.021979: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.688004\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.272\n",
            "iter =       10/      75, loss_segmentation = 0.575 loss_D_trg_fake = 0.846 loss_D = 0.595 loss_seg_trg = 0.000\n",
            "epoch 11, lr 0.021674: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.696759\n",
            "start val!\n",
            "precision per pixel for test: 0.630\n",
            "mIoU for validation: 0.250\n",
            "iter =       11/      75, loss_segmentation = 0.787 loss_D_trg_fake = 0.829 loss_D = 0.646 loss_seg_trg = 0.000\n",
            "epoch 12, lr 0.021369: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.695333\n",
            "start val!\n",
            "precision per pixel for test: 0.663\n",
            "mIoU for validation: 0.237\n",
            "iter =       12/      75, loss_segmentation = 0.606 loss_D_trg_fake = 0.837 loss_D = 0.632 loss_seg_trg = 0.000\n",
            "epoch 13, lr 0.021064: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.634882\n",
            "start val!\n",
            "precision per pixel for test: 0.690\n",
            "mIoU for validation: 0.265\n",
            "iter =       13/      75, loss_segmentation = 0.526 loss_D_trg_fake = 0.730 loss_D = 0.701 loss_seg_trg = 0.000\n",
            "epoch 14, lr 0.020758: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.539774\n",
            "start val!\n",
            "precision per pixel for test: 0.696\n",
            "mIoU for validation: 0.270\n",
            "iter =       14/      75, loss_segmentation = 0.513 loss_D_trg_fake = 0.878 loss_D = 0.617 loss_seg_trg = 0.000\n",
            "epoch 15, lr 0.020451: 100% 1000/1000 [02:22<00:00,  7.02it/s]\n",
            "average loss for train : 0.510180\n",
            "start val!\n",
            "precision per pixel for test: 0.677\n",
            "mIoU for validation: 0.273\n",
            "iter =       15/      75, loss_segmentation = 0.467 loss_D_trg_fake = 0.933 loss_D = 0.604 loss_seg_trg = 0.000\n",
            "epoch 16, lr 0.020144: 100% 1000/1000 [02:22<00:00,  7.02it/s]\n",
            "average loss for train : 0.479770\n",
            "start val!\n",
            "precision per pixel for test: 0.684\n",
            "mIoU for validation: 0.270\n",
            "iter =       16/      75, loss_segmentation = 0.482 loss_D_trg_fake = 0.978 loss_D = 0.619 loss_seg_trg = 0.000\n",
            "epoch 17, lr 0.019837: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.475806\n",
            "start val!\n",
            "precision per pixel for test: 0.694\n",
            "mIoU for validation: 0.285\n",
            "iter =       17/      75, loss_segmentation = 0.529 loss_D_trg_fake = 0.894 loss_D = 0.619 loss_seg_trg = 0.000\n",
            "epoch 18, lr 0.019529: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.443782\n",
            "start val!\n",
            "precision per pixel for test: 0.671\n",
            "mIoU for validation: 0.264\n",
            "iter =       18/      75, loss_segmentation = 0.447 loss_D_trg_fake = 1.038 loss_D = 0.545 loss_seg_trg = 0.000\n",
            "epoch 19, lr 0.019220: 100% 1000/1000 [02:21<00:00,  7.04it/s]\n",
            "average loss for train : 0.427301\n",
            "start val!\n",
            "precision per pixel for test: 0.691\n",
            "mIoU for validation: 0.279\n",
            "iter =       19/      75, loss_segmentation = 0.397 loss_D_trg_fake = 1.058 loss_D = 0.576 loss_seg_trg = 0.000\n",
            "epoch 20, lr 0.018911: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.415260\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.281\n",
            "iter =       20/      75, loss_segmentation = 0.438 loss_D_trg_fake = 1.047 loss_D = 0.601 loss_seg_trg = 0.000\n",
            "epoch 21, lr 0.018601: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.402719\n",
            "start val!\n",
            "precision per pixel for test: 0.702\n",
            "mIoU for validation: 0.284\n",
            "iter =       21/      75, loss_segmentation = 0.367 loss_D_trg_fake = 0.922 loss_D = 0.589 loss_seg_trg = 0.000\n",
            "epoch 22, lr 0.018291: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.392588\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.283\n",
            "iter =       22/      75, loss_segmentation = 0.454 loss_D_trg_fake = 0.833 loss_D = 0.614 loss_seg_trg = 0.000\n",
            "epoch 23, lr 0.017980: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.384351\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.282\n",
            "iter =       23/      75, loss_segmentation = 0.287 loss_D_trg_fake = 1.076 loss_D = 0.582 loss_seg_trg = 0.000\n",
            "epoch 24, lr 0.017668: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.488332\n",
            "start val!\n",
            "precision per pixel for test: 0.660\n",
            "mIoU for validation: 0.265\n",
            "iter =       24/      75, loss_segmentation = 0.360 loss_D_trg_fake = 1.057 loss_D = 0.577 loss_seg_trg = 0.000\n",
            "epoch 25, lr 0.017356: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.418387\n",
            "start val!\n",
            "precision per pixel for test: 0.681\n",
            "mIoU for validation: 0.288\n",
            "iter =       25/      75, loss_segmentation = 0.463 loss_D_trg_fake = 1.154 loss_D = 0.640 loss_seg_trg = 0.000\n",
            "epoch 26, lr 0.017044: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.383576\n",
            "start val!\n",
            "precision per pixel for test: 0.677\n",
            "mIoU for validation: 0.279\n",
            "iter =       26/      75, loss_segmentation = 0.382 loss_D_trg_fake = 1.016 loss_D = 0.587 loss_seg_trg = 0.000\n",
            "epoch 27, lr 0.016730: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.369170\n",
            "start val!\n",
            "precision per pixel for test: 0.690\n",
            "mIoU for validation: 0.282\n",
            "iter =       27/      75, loss_segmentation = 0.314 loss_D_trg_fake = 1.372 loss_D = 0.499 loss_seg_trg = 0.000\n",
            "epoch 28, lr 0.016416: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.362728\n",
            "start val!\n",
            "precision per pixel for test: 0.687\n",
            "mIoU for validation: 0.287\n",
            "iter =       28/      75, loss_segmentation = 0.361 loss_D_trg_fake = 1.474 loss_D = 0.557 loss_seg_trg = 0.000\n",
            "epoch 29, lr 0.016102: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.358965\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.287\n",
            "iter =       29/      75, loss_segmentation = 0.486 loss_D_trg_fake = 1.034 loss_D = 0.606 loss_seg_trg = 0.000\n",
            "epoch 30, lr 0.015786: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.346906\n",
            "start val!\n",
            "precision per pixel for test: 0.693\n",
            "mIoU for validation: 0.283\n",
            "iter =       30/      75, loss_segmentation = 0.286 loss_D_trg_fake = 1.132 loss_D = 0.564 loss_seg_trg = 0.000\n",
            "epoch 31, lr 0.015470: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.342027\n",
            "start val!\n",
            "precision per pixel for test: 0.704\n",
            "mIoU for validation: 0.288\n",
            "iter =       31/      75, loss_segmentation = 0.358 loss_D_trg_fake = 1.175 loss_D = 0.624 loss_seg_trg = 0.000\n",
            "epoch 32, lr 0.015153: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.334573\n",
            "start val!\n",
            "precision per pixel for test: 0.698\n",
            "mIoU for validation: 0.287\n",
            "iter =       32/      75, loss_segmentation = 0.347 loss_D_trg_fake = 0.933 loss_D = 0.627 loss_seg_trg = 0.000\n",
            "epoch 33, lr 0.014836: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.328872\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.292\n",
            "iter =       33/      75, loss_segmentation = 0.328 loss_D_trg_fake = 1.036 loss_D = 0.575 loss_seg_trg = 0.000\n",
            "epoch 34, lr 0.014517: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.327021\n",
            "start val!\n",
            "precision per pixel for test: 0.712\n",
            "mIoU for validation: 0.292\n",
            "iter =       34/      75, loss_segmentation = 0.265 loss_D_trg_fake = 1.320 loss_D = 0.559 loss_seg_trg = 0.000\n",
            "epoch 35, lr 0.014198: 100% 1000/1000 [02:22<00:00,  7.00it/s]\n",
            "average loss for train : 0.324090\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.291\n",
            "iter =       35/      75, loss_segmentation = 0.359 loss_D_trg_fake = 0.972 loss_D = 0.633 loss_seg_trg = 0.000\n",
            "epoch 36, lr 0.013879: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.320762\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.291\n",
            "iter =       36/      75, loss_segmentation = 0.298 loss_D_trg_fake = 1.091 loss_D = 0.558 loss_seg_trg = 0.000\n",
            "epoch 37, lr 0.013558: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.318594\n",
            "start val!\n",
            "precision per pixel for test: 0.697\n",
            "mIoU for validation: 0.281\n",
            "iter =       37/      75, loss_segmentation = 0.454 loss_D_trg_fake = 1.129 loss_D = 0.555 loss_seg_trg = 0.000\n",
            "epoch 38, lr 0.013236: 100% 1000/1000 [02:21<00:00,  7.06it/s]\n",
            "average loss for train : 0.312977\n",
            "start val!\n",
            "precision per pixel for test: 0.713\n",
            "mIoU for validation: 0.293\n",
            "iter =       38/      75, loss_segmentation = 0.285 loss_D_trg_fake = 1.076 loss_D = 0.647 loss_seg_trg = 0.000\n",
            "epoch 39, lr 0.012914: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.312616\n",
            "start val!\n",
            "precision per pixel for test: 0.705\n",
            "mIoU for validation: 0.285\n",
            "iter =       39/      75, loss_segmentation = 0.260 loss_D_trg_fake = 1.031 loss_D = 0.571 loss_seg_trg = 0.000\n",
            "epoch 40, lr 0.012591: 100% 1000/1000 [02:21<00:00,  7.04it/s]\n",
            "average loss for train : 0.307193\n",
            "start val!\n",
            "precision per pixel for test: 0.723\n",
            "mIoU for validation: 0.299\n",
            "iter =       40/      75, loss_segmentation = 0.314 loss_D_trg_fake = 1.079 loss_D = 0.612 loss_seg_trg = 0.000\n",
            "epoch 41, lr 0.012266: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.305747\n",
            "start val!\n",
            "precision per pixel for test: 0.707\n",
            "mIoU for validation: 0.288\n",
            "iter =       41/      75, loss_segmentation = 0.346 loss_D_trg_fake = 1.053 loss_D = 0.601 loss_seg_trg = 0.000\n",
            "epoch 42, lr 0.011941: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.303103\n",
            "start val!\n",
            "precision per pixel for test: 0.714\n",
            "mIoU for validation: 0.293\n",
            "iter =       42/      75, loss_segmentation = 0.219 loss_D_trg_fake = 1.068 loss_D = 0.587 loss_seg_trg = 0.000\n",
            "epoch 43, lr 0.011615: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.299823\n",
            "start val!\n",
            "precision per pixel for test: 0.708\n",
            "mIoU for validation: 0.289\n",
            "iter =       43/      75, loss_segmentation = 0.314 loss_D_trg_fake = 0.933 loss_D = 0.579 loss_seg_trg = 0.000\n",
            "epoch 44, lr 0.011288: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.297449\n",
            "start val!\n",
            "precision per pixel for test: 0.717\n",
            "mIoU for validation: 0.291\n",
            "iter =       44/      75, loss_segmentation = 0.289 loss_D_trg_fake = 0.805 loss_D = 0.651 loss_seg_trg = 0.000\n",
            "epoch 45, lr 0.010960: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.295875\n",
            "start val!\n",
            "precision per pixel for test: 0.709\n",
            "mIoU for validation: 0.286\n",
            "iter =       45/      75, loss_segmentation = 0.292 loss_D_trg_fake = 0.996 loss_D = 0.589 loss_seg_trg = 0.000\n",
            "epoch 46, lr 0.010630: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.294256\n",
            "start val!\n",
            "precision per pixel for test: 0.704\n",
            "mIoU for validation: 0.285\n",
            "iter =       46/      75, loss_segmentation = 0.367 loss_D_trg_fake = 1.039 loss_D = 0.595 loss_seg_trg = 0.000\n",
            "epoch 47, lr 0.010300: 100% 1000/1000 [02:22<00:00,  7.04it/s]\n",
            "average loss for train : 0.292280\n",
            "start val!\n",
            "precision per pixel for test: 0.711\n",
            "mIoU for validation: 0.289\n",
            "iter =       47/      75, loss_segmentation = 0.331 loss_D_trg_fake = 1.151 loss_D = 0.584 loss_seg_trg = 0.000\n",
            "epoch 48, lr 0.009968: 100% 1000/1000 [02:21<00:00,  7.05it/s]\n",
            "average loss for train : 0.289441\n",
            "start val!\n",
            "precision per pixel for test: 0.718\n",
            "mIoU for validation: 0.293\n",
            "iter =       48/      75, loss_segmentation = 0.335 loss_D_trg_fake = 0.936 loss_D = 0.608 loss_seg_trg = 0.000\n",
            "epoch 49, lr 0.009635: 100% 1000/1000 [02:22<00:00,  7.03it/s]\n",
            "average loss for train : 0.289426\n",
            "start val!\n",
            "precision per pixel for test: 0.715\n",
            "mIoU for validation: 0.291\n",
            "iter =       49/      75, loss_segmentation = 0.349 loss_D_trg_fake = 1.024 loss_D = 0.591 loss_seg_trg = 0.000\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5634f90c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5635768c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "[0.9        0.9        0.9        0.75462449 0.65913272 0.65495628\n",
            " 0.64151496 0.63383222 0.9        0.9        0.9        0.89620709\n",
            " 0.48784381 0.9        0.86034131 0.76195598 0.55701935 0.63036108\n",
            " 0.        ]\n",
            "epoch 50, lr 0.009301: 100% 1000/1000 [02:34<00:00,  6.49it/s]\n",
            "average loss for train : 0.491721\n",
            "start val!\n",
            "precision per pixel for test: 0.629\n",
            "mIoU for validation: 0.253\n",
            "iter =       50/      75, loss_segmentation = 0.545 loss_D_trg_fake = 1.110 loss_D = 0.780 loss_seg_trg = 0.284\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5634f90c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5635768c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "[0.9        0.81844932 0.9        0.66564006 0.63034099 0.65303946\n",
            " 0.64815336 0.62149256 0.88873267 0.86061507 0.9        0.9\n",
            " 0.35416207 0.9        0.75431412 0.77670074 0.         0.57263213\n",
            " 0.        ]\n",
            "epoch 51, lr 0.008966: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.518054\n",
            "start val!\n",
            "precision per pixel for test: 0.544\n",
            "mIoU for validation: 0.210\n",
            "iter =       51/      75, loss_segmentation = 0.358 loss_D_trg_fake = 1.244 loss_D = 0.537 loss_seg_trg = 0.199\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5634f90c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "tcmalloc: large alloc 2097152000 bytes == 0x5635768c6000 @  0x7f543febf001 0x7f533ed1954f 0x7f533ed69b58 0x7f533ed6db17 0x7f533ee0c203 0x563431d8b390 0x563431d8b120 0x563431dff679 0x563431dfa02f 0x563431d8caba 0x563431dfbcd4 0x563431d8c9da 0x563431dfaeae 0x563431dfa02f 0x563431df9d43 0x563431ec4302 0x563431ec467d 0x563431ec4526 0x563431e9c1d3 0x563431e9be7c 0x7f543eca7bf7 0x563431e9bd5a\n",
            "[0.9        0.82032007 0.9        0.67417717 0.62148124 0.65706915\n",
            " 0.59701216 0.50947446 0.9        0.89328194 0.9        0.9\n",
            " 0.32626346 0.9        0.83284336 0.88683176 0.2497291  0.46659011\n",
            " 0.        ]\n",
            "epoch 52, lr 0.008629: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.550770\n",
            "start val!\n",
            "precision per pixel for test: 0.434\n",
            "mIoU for validation: 0.193\n",
            "iter =       52/      75, loss_segmentation = 0.367 loss_D_trg_fake = 1.216 loss_D = 0.584 loss_seg_trg = 0.327\n",
            "[0.83374536 0.81170887 0.9        0.68110943 0.61509192 0.65530825\n",
            " 0.59968591 0.49652147 0.9        0.9        0.9        0.9\n",
            " 0.2816565  0.9        0.79463929 0.81114614 0.         0.5990932\n",
            " 0.        ]\n",
            "epoch 53, lr 0.008290: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.562242\n",
            "start val!\n",
            "precision per pixel for test: 0.370\n",
            "mIoU for validation: 0.164\n",
            "iter =       53/      75, loss_segmentation = 0.469 loss_D_trg_fake = 2.147 loss_D = 0.310 loss_seg_trg = 0.160\n",
            "[0.77956301 0.82662827 0.9        0.74193019 0.70585626 0.67450517\n",
            " 0.58154666 0.46991211 0.9        0.9        0.9        0.9\n",
            " 0.20451711 0.9        0.85682184 0.9        0.37708503 0.55439848\n",
            " 0.        ]\n",
            "epoch 54, lr 0.007950: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.436885\n",
            "start val!\n",
            "precision per pixel for test: 0.365\n",
            "mIoU for validation: 0.152\n",
            "iter =       54/      75, loss_segmentation = 0.618 loss_D_trg_fake = 2.029 loss_D = 0.289 loss_seg_trg = 0.149\n",
            "[0.74108303 0.85603744 0.9        0.78026879 0.75660068 0.70937324\n",
            " 0.61686057 0.58322984 0.9        0.9        0.9        0.9\n",
            " 0.2192053  0.9        0.9        0.9        0.3311803  0.5891012\n",
            " 0.        ]\n",
            "epoch 55, lr 0.007609: 100% 1000/1000 [02:34<00:00,  6.49it/s]\n",
            "average loss for train : 0.400689\n",
            "start val!\n",
            "precision per pixel for test: 0.384\n",
            "mIoU for validation: 0.169\n",
            "iter =       55/      75, loss_segmentation = 0.358 loss_D_trg_fake = 2.211 loss_D = 0.267 loss_seg_trg = 0.136\n",
            "[0.76123923 0.9        0.9        0.83032298 0.77341533 0.7479769\n",
            " 0.65306193 0.61205024 0.9        0.9        0.9        0.9\n",
            " 0.27365884 0.9        0.9        0.9        0.         0.58771795\n",
            " 0.        ]\n",
            "epoch 56, lr 0.007265: 100% 1000/1000 [02:34<00:00,  6.46it/s]\n",
            "average loss for train : 0.392320\n",
            "start val!\n",
            "precision per pixel for test: 0.387\n",
            "mIoU for validation: 0.174\n",
            "iter =       56/      75, loss_segmentation = 0.410 loss_D_trg_fake = 2.220 loss_D = 0.399 loss_seg_trg = 0.122\n",
            "[0.79248911 0.9        0.9        0.82803452 0.83182025 0.76207495\n",
            " 0.64805257 0.59226513 0.9        0.9        0.9        0.9\n",
            " 0.39926052 0.9        0.9        0.9        0.         0.65116161\n",
            " 0.        ]\n",
            "epoch 57, lr 0.006920: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.377944\n",
            "start val!\n",
            "precision per pixel for test: 0.382\n",
            "mIoU for validation: 0.171\n",
            "iter =       57/      75, loss_segmentation = 0.288 loss_D_trg_fake = 1.966 loss_D = 0.417 loss_seg_trg = 0.096\n",
            "[0.81749219 0.9        0.9        0.82468104 0.83509684 0.76289099\n",
            " 0.65464735 0.59611201 0.9        0.9        0.9        0.9\n",
            " 0.30528584 0.9        0.9        0.9        0.         0.59700882\n",
            " 0.        ]\n",
            "epoch 58, lr 0.006573: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.378415\n",
            "start val!\n",
            "precision per pixel for test: 0.377\n",
            "mIoU for validation: 0.148\n",
            "iter =       58/      75, loss_segmentation = 0.324 loss_D_trg_fake = 2.457 loss_D = 0.264 loss_seg_trg = 0.179\n",
            "[0.83062464 0.9        0.9        0.85418206 0.85593623 0.84442157\n",
            " 0.70729029 0.62817585 0.9        0.9        0.9        0.9\n",
            " 0.28705162 0.9        0.9        0.9        0.         0.79972047\n",
            " 0.        ]\n",
            "epoch 59, lr 0.006224: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.385124\n",
            "start val!\n",
            "precision per pixel for test: 0.360\n",
            "mIoU for validation: 0.141\n",
            "iter =       59/      75, loss_segmentation = 0.456 loss_D_trg_fake = 1.835 loss_D = 0.438 loss_seg_trg = 0.119\n",
            "[0.89913678 0.9        0.9        0.81925035 0.84578168 0.81508952\n",
            " 0.69892496 0.63885546 0.9        0.9        0.9        0.9\n",
            " 0.33793968 0.9        0.9        0.9        0.         0.68119586\n",
            " 0.        ]\n",
            "epoch 60, lr 0.005873: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.376827\n",
            "start val!\n",
            "precision per pixel for test: 0.347\n",
            "mIoU for validation: 0.129\n",
            "iter =       60/      75, loss_segmentation = 0.353 loss_D_trg_fake = 2.241 loss_D = 0.253 loss_seg_trg = 0.172\n",
            "[0.87409496 0.9        0.9        0.85794407 0.81207824 0.81795698\n",
            " 0.7377339  0.54378062 0.9        0.9        0.9        0.9\n",
            " 0.33677414 0.9        0.9        0.9        0.         0.65093833\n",
            " 0.        ]\n",
            "epoch 61, lr 0.005520: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.372572\n",
            "start val!\n",
            "precision per pixel for test: 0.343\n",
            "mIoU for validation: 0.131\n",
            "iter =       61/      75, loss_segmentation = 0.374 loss_D_trg_fake = 2.792 loss_D = 0.292 loss_seg_trg = 0.144\n",
            "[0.84105402 0.9        0.9        0.84394568 0.85731757 0.87584519\n",
            " 0.77030998 0.61306202 0.9        0.9        0.9        0.9\n",
            " 0.3218078  0.9        0.9        0.9        0.         0.74382883\n",
            " 0.        ]\n",
            "epoch 62, lr 0.005163: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.361942\n",
            "start val!\n",
            "precision per pixel for test: 0.354\n",
            "mIoU for validation: 0.132\n",
            "iter =       62/      75, loss_segmentation = 0.357 loss_D_trg_fake = 1.911 loss_D = 0.367 loss_seg_trg = 0.757\n",
            "[0.85949796 0.9        0.9        0.9        0.87945741 0.84381467\n",
            " 0.78112596 0.57873553 0.9        0.9        0.9        0.9\n",
            " 0.38405874 0.9        0.9        0.9        0.         0.59077471\n",
            " 0.        ]\n",
            "epoch 63, lr 0.004804: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.360184\n",
            "start val!\n",
            "precision per pixel for test: 0.377\n",
            "mIoU for validation: 0.147\n",
            "iter =       63/      75, loss_segmentation = 0.375 loss_D_trg_fake = 2.792 loss_D = 0.221 loss_seg_trg = 0.144\n",
            "[0.8904382  0.9        0.9        0.9        0.9        0.85907799\n",
            " 0.76300019 0.56353998 0.9        0.9        0.9        0.9\n",
            " 0.35500056 0.9        0.9        0.9        0.21366678 0.63630569\n",
            " 0.        ]\n",
            "epoch 64, lr 0.004443: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.362964\n",
            "start val!\n",
            "precision per pixel for test: 0.406\n",
            "mIoU for validation: 0.163\n",
            "iter =       64/      75, loss_segmentation = 0.382 loss_D_trg_fake = 2.609 loss_D = 0.386 loss_seg_trg = 0.178\n",
            "[0.88805437 0.9        0.9        0.9        0.9        0.82789767\n",
            " 0.75550401 0.64575487 0.9        0.9        0.9        0.9\n",
            " 0.34539455 0.9        0.76250237 0.9        0.21671693 0.63522589\n",
            " 0.        ]\n",
            "epoch 65, lr 0.004077: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.358011\n",
            "start val!\n",
            "precision per pixel for test: 0.414\n",
            "mIoU for validation: 0.175\n",
            "iter =       65/      75, loss_segmentation = 0.469 loss_D_trg_fake = 2.277 loss_D = 0.351 loss_seg_trg = 0.076\n",
            "[0.9        0.9        0.9        0.9        0.9        0.87406069\n",
            " 0.70892149 0.67756951 0.9        0.9        0.9        0.9\n",
            " 0.39211014 0.9        0.88645923 0.9        0.17914686 0.71544528\n",
            " 0.        ]\n",
            "epoch 66, lr 0.003709: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.348116\n",
            "start val!\n",
            "precision per pixel for test: 0.420\n",
            "mIoU for validation: 0.175\n",
            "iter =       66/      75, loss_segmentation = 0.373 loss_D_trg_fake = 1.992 loss_D = 0.372 loss_seg_trg = 0.186\n",
            "[0.9        0.9        0.9        0.9        0.9        0.86049485\n",
            " 0.72750819 0.69535029 0.9        0.9        0.9        0.9\n",
            " 0.37040189 0.9        0.9        0.9        0.         0.68706638\n",
            " 0.        ]\n",
            "epoch 67, lr 0.003336: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.341837\n",
            "start val!\n",
            "precision per pixel for test: 0.407\n",
            "mIoU for validation: 0.173\n",
            "iter =       67/      75, loss_segmentation = 0.339 loss_D_trg_fake = 1.475 loss_D = 0.398 loss_seg_trg = 0.063\n",
            "[0.9        0.9        0.9        0.9        0.9        0.87737226\n",
            " 0.75204778 0.70158809 0.9        0.9        0.9        0.9\n",
            " 0.35226625 0.9        0.9        0.9        0.         0.69578725\n",
            " 0.        ]\n",
            "epoch 68, lr 0.002958: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.335567\n",
            "start val!\n",
            "precision per pixel for test: 0.397\n",
            "mIoU for validation: 0.163\n",
            "iter =       68/      75, loss_segmentation = 0.437 loss_D_trg_fake = 2.465 loss_D = 0.465 loss_seg_trg = 0.046\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.79613161 0.75349754 0.9        0.9        0.9        0.9\n",
            " 0.42556441 0.9        0.9        0.9        0.         0.74687016\n",
            " 0.        ]\n",
            "epoch 69, lr 0.002575: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.331612\n",
            "start val!\n",
            "precision per pixel for test: 0.397\n",
            "mIoU for validation: 0.161\n",
            "iter =       69/      75, loss_segmentation = 0.310 loss_D_trg_fake = 1.649 loss_D = 0.415 loss_seg_trg = 0.099\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.75691849 0.72544998 0.9        0.9        0.9        0.9\n",
            " 0.40744245 0.9        0.9        0.9        0.         0.76731288\n",
            " 0.        ]\n",
            "epoch 70, lr 0.002185: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.331691\n",
            "start val!\n",
            "precision per pixel for test: 0.396\n",
            "mIoU for validation: 0.160\n",
            "iter =       70/      75, loss_segmentation = 0.310 loss_D_trg_fake = 1.892 loss_D = 0.346 loss_seg_trg = 0.175\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.74228513 0.71718091 0.9        0.9        0.9        0.9\n",
            " 0.43138999 0.9        0.9        0.9        0.         0.7519092\n",
            " 0.        ]\n",
            "epoch 71, lr 0.001787: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.332126\n",
            "start val!\n",
            "precision per pixel for test: 0.384\n",
            "mIoU for validation: 0.155\n",
            "iter =       71/      75, loss_segmentation = 0.356 loss_D_trg_fake = 2.048 loss_D = 0.377 loss_seg_trg = 0.060\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.7418282  0.76473385 0.9        0.9        0.9        0.9\n",
            " 0.43469188 0.9        0.9        0.9        0.         0.8386364\n",
            " 0.        ]\n",
            "epoch 72, lr 0.001380: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.327518\n",
            "start val!\n",
            "precision per pixel for test: 0.381\n",
            "mIoU for validation: 0.155\n",
            "iter =       72/      75, loss_segmentation = 0.380 loss_D_trg_fake = 1.911 loss_D = 0.363 loss_seg_trg = 0.074\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.75367481 0.77556247 0.9        0.9        0.9        0.9\n",
            " 0.47183552 0.9        0.9        0.9        0.         0.79003054\n",
            " 0.        ]\n",
            "epoch 73, lr 0.000958: 100% 1000/1000 [02:34<00:00,  6.48it/s]\n",
            "average loss for train : 0.326862\n",
            "start val!\n",
            "precision per pixel for test: 0.369\n",
            "mIoU for validation: 0.148\n",
            "iter =       73/      75, loss_segmentation = 0.337 loss_D_trg_fake = 2.048 loss_D = 0.417 loss_seg_trg = 0.153\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.76302242 0.80402058 0.9        0.9        0.9        0.9\n",
            " 0.41342744 0.9        0.9        0.9        0.         0.81604326\n",
            " 0.        ]\n",
            "epoch 74, lr 0.000513: 100% 1000/1000 [02:34<00:00,  6.47it/s]\n",
            "average loss for train : 0.329565\n",
            "start val!\n",
            "precision per pixel for test: 0.375\n",
            "mIoU for validation: 0.148\n",
            "iter =       74/      75, loss_segmentation = 0.295 loss_D_trg_fake = 2.093 loss_D = 0.353 loss_seg_trg = 0.086\n",
            "[0.9        0.9        0.9        0.9        0.9        0.9\n",
            " 0.79164928 0.81252176 0.9        0.9        0.9        0.9\n",
            " 0.39266318 0.9        0.9        0.9        0.         0.82122779\n",
            " 0.        ]\n"
          ]
        }
      ],
      "source": [
        "#c) Pseudo Labelling of target domain as in BDL\n",
        "!python trainGTAtoCityscapesSSL.py"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_rLl-x-D1Dx9",
        "outputId": "b72e60a3-cfdf-4ba6-b402-bda60ce05d14"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "load model from /content/drive/MyDrive/checkpoints_101_sgd_da_DSC_withpseudolabels/best_crossentropy_loss.pth ...\n",
            "The best epoch is 40\n",
            "Done!\n",
            "start test!\n",
            "test: 100% 250/250 [05:29<00:00,  1.32s/it]\n",
            "IoU for class road: 0.863\n",
            "IoU for class sidewalk: 0.312\n",
            "IoU for class building: 0.760\n",
            "IoU for class wall: 0.193\n",
            "IoU for class fence: 0.074\n",
            "IoU for class pole: 0.205\n",
            "IoU for class light: 0.151\n",
            "IoU for class sign: 0.030\n",
            "IoU for class vegetation: 0.808\n",
            "IoU for class terrain: 0.219\n",
            "IoU for class sky: 0.730\n",
            "IoU for class person: 0.369\n",
            "IoU for class rider: 0.036\n",
            "IoU for class car: 0.705\n",
            "IoU for class truck: 0.152\n",
            "IoU for class bus: 0.032\n",
            "IoU for class train: 0.002\n",
            "IoU for class motocycle: 0.037\n",
            "IoU for class bicycle: 0.000\n",
            "precision per pixel for test: 0.723\n",
            "mIoU for validation: 0.299\n"
          ]
        }
      ],
      "source": [
        "#c) Pseudo Labelling of target domain as in BDL\n",
        "!python eval.py"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "machine_shape": "hm",
      "name": "Exam.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}