Vgg_16_transfer_learning_Normalization.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/RuthKassahun/Melanoma_Classification/blob/main/Vgg_16_transfer_learning_Normalization.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "itSCyw3ipYfv"
      },
      "outputs": [],
      "source": [
        "!mkdir -p ~/.kaggle\n",
        "!cp kaggle.json ~/.kaggle/"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!kaggle datasets download -d salader/dogs-vs-cats"
      ],
      "metadata": {
        "id": "KW42KSn9pfyA",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "65abb533-1cdd-440e-aa3c-eb3c16021b89"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Warning: Your Kaggle API key is readable by other users on this system! To fix this, you can run 'chmod 600 /root/.kaggle/kaggle.json'\n",
            "Downloading dogs-vs-cats.zip to /content\n",
            "100% 1.06G/1.06G [00:05<00:00, 196MB/s]\n",
            "100% 1.06G/1.06G [00:05<00:00, 197MB/s]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import zipfile\n",
        "zip_ref = zipfile.ZipFile('/content/dogs-vs-cats.zip', 'r')\n",
        "zip_ref.extractall('/content')\n",
        "zip_ref.close()"
      ],
      "metadata": {
        "id": "CXz412Aip5B4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow\n",
        "from tensorflow import keras\n",
        "from keras import Sequential\n",
        "from keras.layers import Dense,Flatten\n",
        "from keras.applications.vgg16 import VGG16"
      ],
      "metadata": {
        "id": "8dAgedWkp8Pi"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ChjnmP4LVv30",
        "outputId": "36931d3d-50d4-4ad6-b956-b5b4c0dd5547"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "conv_base = VGG16(\n",
        "    weights='imagenet',\n",
        "    include_top = False,\n",
        "    input_shape=(150,150,3)\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xGeyBIoxp_ji",
        "outputId": "baa55ad8-e3ec-478a-f9ed-069abc013701"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "58889256/58889256 [==============================] - 1s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "conv_base.trainable = True\n",
        "\n",
        "set_trainable = False\n",
        "\n",
        "for layer in conv_base.layers:\n",
        "  if layer.name == 'block5_conv1':\n",
        "    set_trainable = True\n",
        "  if set_trainable:\n",
        "    layer.trainable = True\n",
        "  else:\n",
        "    layer.trainable = False\n",
        "\n",
        "for layer in conv_base.layers:\n",
        "  print(layer.name,layer.trainable)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rsRlwtCUqB81",
        "outputId": "219389d7-b7d1-4194-9927-4f78da90af80"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "input_1 False\n",
            "block1_conv1 False\n",
            "block1_conv2 False\n",
            "block1_pool False\n",
            "block2_conv1 False\n",
            "block2_conv2 False\n",
            "block2_pool False\n",
            "block3_conv1 False\n",
            "block3_conv2 False\n",
            "block3_conv3 False\n",
            "block3_pool False\n",
            "block4_conv1 False\n",
            "block4_conv2 False\n",
            "block4_conv3 False\n",
            "block4_pool False\n",
            "block5_conv1 True\n",
            "block5_conv2 True\n",
            "block5_conv3 True\n",
            "block5_pool True\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "conv_base.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qMNnxAe8qxaj",
        "outputId": "1f2a9589-596f-48ce-ae25-b4f768f10bc7"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"vgg16\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 150, 150, 3)]     0         \n",
            "                                                                 \n",
            " block1_conv1 (Conv2D)       (None, 150, 150, 64)      1792      \n",
            "                                                                 \n",
            " block1_conv2 (Conv2D)       (None, 150, 150, 64)      36928     \n",
            "                                                                 \n",
            " block1_pool (MaxPooling2D)  (None, 75, 75, 64)        0         \n",
            "                                                                 \n",
            " block2_conv1 (Conv2D)       (None, 75, 75, 128)       73856     \n",
            "                                                                 \n",
            " block2_conv2 (Conv2D)       (None, 75, 75, 128)       147584    \n",
            "                                                                 \n",
            " block2_pool (MaxPooling2D)  (None, 37, 37, 128)       0         \n",
            "                                                                 \n",
            " block3_conv1 (Conv2D)       (None, 37, 37, 256)       295168    \n",
            "                                                                 \n",
            " block3_conv2 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_conv3 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_pool (MaxPooling2D)  (None, 18, 18, 256)       0         \n",
            "                                                                 \n",
            " block4_conv1 (Conv2D)       (None, 18, 18, 512)       1180160   \n",
            "                                                                 \n",
            " block4_conv2 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_conv3 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_pool (MaxPooling2D)  (None, 9, 9, 512)         0         \n",
            "                                                                 \n",
            " block5_conv1 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv2 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv3 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_pool (MaxPooling2D)  (None, 4, 4, 512)         0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,714,688\n",
            "Trainable params: 7,079,424\n",
            "Non-trainable params: 7,635,264\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model = Sequential()\n",
        "\n",
        "model.add(conv_base)\n",
        "model.add(Flatten())\n",
        "model.add(Dense(256,activation='relu'))\n",
        "model.add(Dense(1,activation='sigmoid'))"
      ],
      "metadata": {
        "id": "MuCh9EqrsMPK"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# generators\n",
        "train_ds = keras.utils.image_dataset_from_directory(\n",
        "    directory = '/content/drive/MyDrive/MultiClass_Skin_lesion/train/',\n",
        "    labels='inferred',\n",
        "    label_mode = 'int',\n",
        "    batch_size=32,\n",
        "    image_size=(150,150)\n",
        ")\n",
        "\n",
        "validation_ds = keras.utils.image_dataset_from_directory(\n",
        "    directory = '/content/drive/MyDrive/MultiClass_Skin_lesion/val/',\n",
        "    labels='inferred',\n",
        "    label_mode = 'int',\n",
        "    batch_size=32,\n",
        "    image_size=(150,150)\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "js3QmFghq1AC",
        "outputId": "d5a5f0ab-29db-4c4d-b6b1-a20e9be51ba7"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 5082 files belonging to 3 classes.\n",
            "Found 1270 files belonging to 3 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Normalize\n",
        "def process(image,label):\n",
        "    image = tensorflow.cast(image/255. ,tensorflow.float32)\n",
        "    return image,label\n",
        "\n",
        "train_ds = train_ds.map(process)\n",
        "validation_ds = validation_ds.map(process)"
      ],
      "metadata": {
        "id": "DjbuBEQkr3DE"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install tensorflow-addons"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FtEE8sj1XJNw",
        "outputId": "86613ced-10a9-4d1f-ed7a-1bdcb8d63eed"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting tensorflow-addons\n",
            "  Downloading tensorflow_addons-0.19.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.1 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.1/1.1 MB\u001b[0m \u001b[31m49.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: typeguard>=2.7 in /usr/local/lib/python3.8/dist-packages (from tensorflow-addons) (2.7.1)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.8/dist-packages (from tensorflow-addons) (21.3)\n",
            "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /usr/local/lib/python3.8/dist-packages (from packaging->tensorflow-addons) (3.0.9)\n",
            "Installing collected packages: tensorflow-addons\n",
            "Successfully installed tensorflow-addons-0.19.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow_addons as tfa"
      ],
      "metadata": {
        "id": "9HnlI2uQYYve"
      },
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(\n",
        "    optimizer=keras.optimizers.RMSprop(lr=1e-5),\n",
        "    loss='binary_crossentropy',\n",
        "    metrics=['accuracy',tfa.metrics.CohenKappa(num_classes=3, sparse_labels=True)]\n",
        "  )"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nxUjDM6Nr5Qp",
        "outputId": "76c6b86f-0688-4bc3-bb9c-7b1cecee48e2"
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.8/dist-packages/keras/optimizers/optimizer_v2/rmsprop.py:135: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  super(RMSprop, self).__init__(name, **kwargs)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "history = model.fit(train_ds,epochs=10,validation_data=validation_ds)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tiat79J_sG6W",
        "outputId": "bb7cbc0e-4f6d-423e-c049-c08c1ca3bbc3"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "159/159 [==============================] - 779s 5s/step - loss: 0.4855 - accuracy: 0.6592 - cohen_kappa: 0.0000e+00 - val_loss: 0.4135 - val_accuracy: 0.7559 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 2/10\n",
            "159/159 [==============================] - 59s 356ms/step - loss: 0.3689 - accuracy: 0.7536 - cohen_kappa: 0.0000e+00 - val_loss: 0.3545 - val_accuracy: 0.7748 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 3/10\n",
            "159/159 [==============================] - 53s 323ms/step - loss: 0.3007 - accuracy: 0.7621 - cohen_kappa: 0.0000e+00 - val_loss: 0.2766 - val_accuracy: 0.7504 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 4/10\n",
            "159/159 [==============================] - 53s 321ms/step - loss: 0.2397 - accuracy: 0.7666 - cohen_kappa: 0.0000e+00 - val_loss: 0.2755 - val_accuracy: 0.7992 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 5/10\n",
            "159/159 [==============================] - 53s 323ms/step - loss: 0.1760 - accuracy: 0.7806 - cohen_kappa: 0.0000e+00 - val_loss: 0.2268 - val_accuracy: 0.7969 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 6/10\n",
            "159/159 [==============================] - 54s 325ms/step - loss: 0.1054 - accuracy: 0.7834 - cohen_kappa: 0.0000e+00 - val_loss: 0.2192 - val_accuracy: 0.7945 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 7/10\n",
            "159/159 [==============================] - 54s 327ms/step - loss: 0.0230 - accuracy: 0.7898 - cohen_kappa: 0.0000e+00 - val_loss: 0.1383 - val_accuracy: 0.7630 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 8/10\n",
            "159/159 [==============================] - 53s 325ms/step - loss: -0.0789 - accuracy: 0.7920 - cohen_kappa: 0.0000e+00 - val_loss: 0.0974 - val_accuracy: 0.7606 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 9/10\n",
            "159/159 [==============================] - 61s 375ms/step - loss: -0.2119 - accuracy: 0.7904 - cohen_kappa: 0.0000e+00 - val_loss: 0.0322 - val_accuracy: 0.7976 - val_cohen_kappa: 0.0000e+00\n",
            "Epoch 10/10\n",
            "159/159 [==============================] - 55s 331ms/step - loss: -0.3846 - accuracy: 0.7926 - cohen_kappa: 0.0000e+00 - val_loss: -0.0261 - val_accuracy: 0.7921 - val_cohen_kappa: 0.0000e+00\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(history.history['accuracy'],color='red',label='train')\n",
        "plt.plot(history.history['val_accuracy'],color='blue',label='validation')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "M8c6eVF4sVTT",
        "outputId": "31cf0992-18d7-45a0-bcb0-65aad3526fa3"
      },
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['loss'],color='red',label='train')\n",
        "plt.plot(history.history['val_loss'],color='blue',label='validation')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "Ap3-7R0swWTo",
        "outputId": "8458af51-aff3-4e4d-8558-e2db72ffdabc"
      },
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "xjaBZM7xiKO7"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}