Demo_Notebook.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Demo Notebook.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "ykgkT8riKZRb"
      },
      "source": [
        "import numpy as np\n",
        "import seaborn as sns\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import cv2\n",
        "import seaborn as sns\n",
        "from keras_preprocessing.image import ImageDataGenerator\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "import albumentations as A\n",
        "from tensorflow.keras.utils import Sequence\n",
        "from sklearn.model_selection import train_test_split\n",
        "import cv2\n",
        "from google.colab.patches import cv2_imshow\n",
        "import skimage.exposure\n",
        "import os"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m6as12EOK_py"
      },
      "source": [
        "!unzip \"/content/drive/MyDrive/mammography_images.zip\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6i7A8FR7LJea"
      },
      "source": [
        "train=pd.read_csv(\"/content/mammography_images/Training_set.csv\")\n",
        "test=pd.read_csv(\"/content/mammography_images/Testing_set.csv\")\n",
        "sub=pd.read_csv(\"/content/mammography_images/sample_submission.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "id": "ewk47YVXMvsQ",
        "outputId": "e0a918bd-9372-4708-df9d-5f0ee9dd9c99"
      },
      "source": [
        "train.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>filename</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Image_1.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Image_2.jpg</td>\n",
              "      <td>Density1Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Image_3.jpg</td>\n",
              "      <td>Density1Malignant</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Image_4.jpg</td>\n",
              "      <td>Density1Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Image_5.jpg</td>\n",
              "      <td>Density1Malignant</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      filename              label\n",
              "0  Image_1.jpg     Density3Benign\n",
              "1  Image_2.jpg     Density1Benign\n",
              "2  Image_3.jpg  Density1Malignant\n",
              "3  Image_4.jpg     Density1Benign\n",
              "4  Image_5.jpg  Density1Malignant"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Uh2yi0HZRouP",
        "outputId": "0cf93589-7195-4211-9f21-677d1dc6e11f"
      },
      "source": [
        "print(\"len of training set\", len(train))\n",
        "print(\"len of testing set\", len(test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "len of training set 5724\n",
            "len of testing set 1908\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1EE-XpALLdc3",
        "outputId": "f94b2b58-3205-424c-9461-c8c84ead3be8"
      },
      "source": [
        "(train[\"label\"].value_counts())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Density2Malignant    1728\n",
              "Density1Malignant    1620\n",
              "Density3Benign        702\n",
              "Density1Benign        648\n",
              "Density3Malignant     432\n",
              "Density4Benign        324\n",
              "Density2Benign        216\n",
              "Density4Malignant      54\n",
              "Name: label, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 342
        },
        "id": "IeyjXGUOLelK",
        "outputId": "b003dc89-4dec-4827-ccc5-06abb6a05c6f"
      },
      "source": [
        "plt.figure(figsize=(12,4))\n",
        "sns.countplot(x=train[\"label\"],data=train,)\n",
        "plt.xticks(rotation=30)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(array([0, 1, 2, 3, 4, 5, 6, 7]), <a list of 8 Text major ticklabel objects>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 402
        },
        "id": "IZvWvLtgTi4d",
        "outputId": "2fd24b66-0c13-48dc-c972-9f4deab34138"
      },
      "source": [
        "train[train[\"label\"]==\"Density3Benign\"]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>filename</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Image_1.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>Image_7.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>Image_20.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24</th>\n",
              "      <td>Image_25.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>33</th>\n",
              "      <td>Image_34.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5690</th>\n",
              "      <td>Image_5691.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5705</th>\n",
              "      <td>Image_5706.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5715</th>\n",
              "      <td>Image_5716.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5717</th>\n",
              "      <td>Image_5718.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5723</th>\n",
              "      <td>Image_5724.jpg</td>\n",
              "      <td>Density3Benign</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>702 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "            filename           label\n",
              "0        Image_1.jpg  Density3Benign\n",
              "6        Image_7.jpg  Density3Benign\n",
              "19      Image_20.jpg  Density3Benign\n",
              "24      Image_25.jpg  Density3Benign\n",
              "33      Image_34.jpg  Density3Benign\n",
              "...              ...             ...\n",
              "5690  Image_5691.jpg  Density3Benign\n",
              "5705  Image_5706.jpg  Density3Benign\n",
              "5715  Image_5716.jpg  Density3Benign\n",
              "5717  Image_5718.jpg  Density3Benign\n",
              "5723  Image_5724.jpg  Density3Benign\n",
              "\n",
              "[702 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Tb1hkqZcpLvA"
      },
      "source": [
        "def preprocess(image):\n",
        "    kernel = np.array([[0,-1,0], [-1,5,-1], [0,-1,0]])\n",
        "    im = cv2.filter2D(image, -1, kernel)\n",
        "    #out2 = skimage.exposure.rescale_intensity(im, in_range=(150,200), out_range=(0,255))\n",
        "    out1=(cv2.normalize(im, (224,224),0, 255, cv2.NORM_MINMAX))\n",
        "    return out1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U6ST57D3Mr3r"
      },
      "source": [
        "X_train, X_valid = train_test_split(train, test_size=0.1,stratify=train[\"label\"],shuffle=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zadg3bosRya9",
        "outputId": "cca11469-f714-4a61-afd3-c14c1a753800"
      },
      "source": [
        "print(\"len of training set\", len(X_train))\n",
        "print(\"len of validation set\", len(X_valid))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "len of training set 5151\n",
            "len of validation set 573\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WSk6zmh5NXZv"
      },
      "source": [
        "datagen=ImageDataGenerator(rescale=1./255,preprocessing_function=preprocess)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HgIPTadsNNp7",
        "outputId": "00fe6813-d88a-49f9-dbd5-b351d282c54a"
      },
      "source": [
        "train_generator=datagen.flow_from_dataframe(\n",
        "dataframe=X_train,\n",
        "directory=\"/content/mammography_images/train\",\n",
        "x_col=\"filename\",\n",
        "y_col=\"label\",\n",
        "batch_size=32,\n",
        "seed=42,\n",
        "shuffle=True,\n",
        "class_mode=\"categorical\",\n",
        "target_size=(224,224))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 5151 validated image filenames belonging to 8 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 425
        },
        "id": "V85HgpY5NVra",
        "outputId": "639b92dc-1e17-459e-87af-2ba2e61be211"
      },
      "source": [
        "x,y = train_generator.next()\n",
        "for i in range(2):\n",
        "    image = x[i]\n",
        "    plt.figure(figsize=(6,8))\n",
        "    plt.subplot(1,2,i+1)\n",
        "    plt.imshow(image)\n",
        "    print(image.shape)\n",
        "   \n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(224, 224, 3)\n",
            "(224, 224, 3)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "m1hA3Yh-QIfs",
        "outputId": "2dd542f4-3603-456a-c8fb-7a2bdcd86027"
      },
      "source": [
        "valid_datagen=ImageDataGenerator(rescale=1./255.,preprocessing_function=preprocess)\n",
        "valid_generator=valid_datagen.flow_from_dataframe( \n",
        "dataframe=X_valid,\n",
        "directory=\"/content/mammography_images/train\",\n",
        "x_col=\"filename\",\n",
        "y_col=\"label\",\n",
        "batch_size=32,\n",
        "seed=42,\n",
        "shuffle=True,\n",
        "class_mode=\"categorical\",\n",
        "target_size=(224,224))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 573 validated image filenames belonging to 8 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OA91e9Q2RmtA"
      },
      "source": [
        "from tensorflow.keras.layers import Dense, Flatten, GlobalAveragePooling2D, BatchNormalization, Dropout,AveragePooling2D\n",
        "from tensorflow.keras.applications.resnet import ResNet50\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.applications import InceptionV3,DenseNet201,EfficientNetB7, MobileNetV2,Xception,VGG16,NASNetMobile\n",
        "from keras.applications.inception_resnet_v2 import InceptionResNetV2\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau\n",
        "from keras.models import Model\n",
        "from keras.models import Sequential\n",
        "from keras.regularizers import *\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bEq2Mh8LR8ke"
      },
      "source": [
        "checkpoint_path = \"training_0/cp.ckpt\"\n",
        "checkpoint_dir = os.path.dirname(checkpoint_path)\n",
        "my_callbacks = [\n",
        "               ModelCheckpoint(checkpoint_path, monitor = 'val_accuracy',verbose = 1,save_weights_only=True, save_best_only = True,mode=\"max\"),\n",
        "               EarlyStopping(monitor='val_loss', patience=5, verbose=0, mode='min'),\n",
        "               ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=5, verbose=1, mode='min',min_delta=1e-4)\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9fBfb4I0SCEI"
      },
      "source": [
        "STEP_SIZE_TRAIN=train_generator.n//train_generator.batch_size\n",
        "STEP_SIZE_VALID=valid_generator.n//valid_generator.batch_size"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z1TU-wYfSO1Z"
      },
      "source": [
        "def build_model():\n",
        "  model = Sequential()\n",
        "\n",
        "  conv_base = DenseNet201(input_shape=(224,224,3), include_top=False, pooling='max',weights='imagenet')\n",
        "  model.add(conv_base)\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(Dense(2048, activation='relu', kernel_regularizer=l1_l2(0.01)))\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(Dense(8, activation='softmax'))\n",
        "\n",
        "  train_layers = [layer for layer in conv_base.layers[::-1][:5]]\n",
        "\n",
        "  for layer in conv_base.layers:\n",
        "    if layer in train_layers:\n",
        "      layer.trainable = True\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jpNf7802SclF",
        "outputId": "2d8df0e7-2591-4ae3-fc33-0a4e2108669d"
      },
      "source": [
        "my_model=build_model()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/densenet/densenet201_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "74842112/74836368 [==============================] - 1s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TI35fmXnSfTF"
      },
      "source": [
        "my_model.compile(optimizer =tf.keras.optimizers.Adam(learning_rate=0.00001,decay=0.0001),metrics=[\"accuracy\"],loss= tf.keras.losses.CategoricalCrossentropy(label_smoothing=0.1))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YIFL0K-PSiwV",
        "outputId": "8afcf30f-78fa-4454-8d81-d89f81bdcb3b"
      },
      "source": [
        "my_model.fit(\n",
        "        train_generator,\n",
        "        steps_per_epoch=STEP_SIZE_TRAIN,\n",
        "        epochs=5,\n",
        "        validation_data=valid_generator,\n",
        "        validation_steps=STEP_SIZE_VALID,callbacks=[my_callbacks])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "160/160 [==============================] - 90s 560ms/step - loss: 0.6533 - accuracy: 1.0000 - val_loss: 0.6591 - val_accuracy: 0.9926\n",
            "\n",
            "Epoch 00001: val_accuracy did not improve from 0.99816\n",
            "Epoch 2/5\n",
            "160/160 [==============================] - 91s 570ms/step - loss: 0.6348 - accuracy: 1.0000 - val_loss: 0.6548 - val_accuracy: 0.9945\n",
            "\n",
            "Epoch 00002: val_accuracy did not improve from 0.99816\n",
            "Epoch 3/5\n",
            "160/160 [==============================] - 92s 576ms/step - loss: 0.6301 - accuracy: 1.0000 - val_loss: 0.6548 - val_accuracy: 0.9945\n",
            "\n",
            "Epoch 00003: val_accuracy did not improve from 0.99816\n",
            "Epoch 4/5\n",
            "160/160 [==============================] - 93s 581ms/step - loss: 0.6170 - accuracy: 1.0000 - val_loss: 0.6355 - val_accuracy: 0.9982\n",
            "\n",
            "Epoch 00004: val_accuracy did not improve from 0.99816\n",
            "Epoch 5/5\n",
            "160/160 [==============================] - 93s 582ms/step - loss: 0.6073 - accuracy: 1.0000 - val_loss: 0.6192 - val_accuracy: 1.0000\n",
            "\n",
            "Epoch 00005: val_accuracy improved from 0.99816 to 1.00000, saving model to training_0/cp.ckpt\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7fc97c13d3d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 76
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mwTjyZZYuC8B",
        "outputId": "a4be8820-7cf4-4d04-84c5-2e74cba3725e"
      },
      "source": [
        "m.evaluate(valid_generator,verbose=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "18/18 [==============================] - 7s 136ms/step - loss: 0.6230 - accuracy: 1.0000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6186696290969849, 1.0]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 97
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_HgBxYOlSlhG",
        "outputId": "6ccb8267-10b1-4cfc-8ba3-f375cf21169b"
      },
      "source": [
        "my_model.load_weights(checkpoint_path)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7fc97c4ca5d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 77
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6w23OzKIT03m",
        "outputId": "499e2678-3812-4589-b247-fc4863636684"
      },
      "source": [
        "my_model.evaluate(valid_generator,verbose=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "18/18 [==============================] - 3s 135ms/step - loss: 0.6187 - accuracy: 1.0000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6186696290969849, 1.0]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 78
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ODaGcA-TT1kV",
        "outputId": "71bac5fe-46ef-449b-a4c9-1763fb765076"
      },
      "source": [
        "train_generator.class_indices"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'Density1Benign': 0,\n",
              " 'Density1Malignant': 1,\n",
              " 'Density2Benign': 2,\n",
              " 'Density2Malignant': 3,\n",
              " 'Density3Benign': 4,\n",
              " 'Density3Malignant': 5,\n",
              " 'Density4Benign': 6,\n",
              " 'Density4Malignant': 7}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 79
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5roV7vhxUaw4",
        "outputId": "7109e40e-c2ac-487f-bc4e-77f03bfb2d6f"
      },
      "source": [
        "test_datagen=ImageDataGenerator(rescale=1./255.,preprocessing_function=preprocess)\n",
        "test_generator=valid_datagen.flow_from_dataframe(\n",
        "dataframe=test,\n",
        "directory=\"/content/mammography_images/test\",\n",
        "x_col=\"filename\",\n",
        "y_col=None,\n",
        "batch_size=1,\n",
        "seed=42,\n",
        "shuffle=False,\n",
        "class_mode=None,\n",
        "target_size=(224,224))\n",
        "STEP_SIZE_TEST=test_generator.n//test_generator.batch_size"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 1908 validated image filenames.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1vrIpTUlU0l6",
        "outputId": "5103ff60-2f6e-485e-c0e4-cc2c9d1f5627"
      },
      "source": [
        "pred1=np.argmax(my_model.predict(test_generator,steps=STEP_SIZE_TEST,verbose=1),axis=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1908/1908 [==============================] - 44s 23ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4UiqOYOrVlw8",
        "outputId": "6bc1446c-048f-46bb-ba71-2dc6377480b6"
      },
      "source": [
        "pred1"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([4, 0, 3, ..., 3, 3, 3])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 87
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "62rCGDnMU5HD"
      },
      "source": [
        "sub[\"label\"]=pred1\n",
        "sub[\"label\"]=sub[\"label\"].map({0:'Density1Benign',\n",
        " 1:'Density1Malignant',\n",
        " 2:'Density2Benign',\n",
        " 3:'Density2Malignant',\n",
        " 4: 'Density3Benign',\n",
        " 5: 'Density3Malignant',\n",
        " 6:'Density4Benign',\n",
        " 7:  'Density4Malignant'})"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e-F3Qhx5VXZ0",
        "outputId": "670bc899-a8d8-4ca5-c457-892ab8a2f451"
      },
      "source": [
        "sub[\"label\"].value_counts()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Density2Malignant    576\n",
              "Density1Malignant    540\n",
              "Density3Benign       234\n",
              "Density1Benign       216\n",
              "Density3Malignant    144\n",
              "Density4Benign       108\n",
              "Density2Benign        72\n",
              "Density4Malignant     18\n",
              "Name: label, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 89
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xJcK-Np9Vd7V"
      },
      "source": [
        "sub.to_csv(\"9.csv\",index=False)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}