VGG19_crop.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "ttZ16wFrjC1p"
      },
      "outputs": [],
      "source": [
        "import math\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib.image as mpimg\n",
        "import pickle\n",
        "import os\n",
        "import tensorflow as tf\n",
        "import sys\n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import *\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from tensorflow.keras import backend as k"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MfLMd4iTlvDq",
        "outputId": "f5081d2b-225e-46b0-9624-10af932a5a7c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "path = '/content/drive/MyDrive/376_project/newcancerdata/'\n",
        "oldpath = '/content/drive/MyDrive/376_project/cancerdata'\n",
        "BATCH_SIZE = 64"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "juKkjPtul2PR",
        "outputId": "69bc6b69-0a46-4bbe-dbba-c8b3fa9af852"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(10000, 3)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<bound method NDFrame.head of        index                                            id label\n",
              "0         17  0003378267bd3d9964d701ff9e52ddce1b027e46.tif     0\n",
              "1         49  000af35befdd9ab2e24fac80fb6508dfd1edd172.tif     1\n",
              "2         66  0011a007b64a4a0f351ee4d7d00f237b1713b9bf.tif     0\n",
              "3        160  002cf4e8faa1b238d67b3066fbeb382d128bc6e9.tif     0\n",
              "4        183  0033eb6f1ae535166311e5837b006cba336e9200.tif     0\n",
              "...      ...                                           ...   ...\n",
              "9995  219929  ffe4b21fe0e2e13c677a0508aaa2717a0cf2ba08.tif     0\n",
              "9996  219957  ffec122459ed1be2cb4a94009c115113a992813d.tif     0\n",
              "9997  219958  ffec4a03e0a1b8d70e2bab767ac0f1803519e71a.tif     1\n",
              "9998  219981  fff2b1244d1d9492ba0950f3221628add0e1cbd4.tif     0\n",
              "9999  219986  fff3cf6f5992363edf773aa539a98679aad19180.tif     0\n",
              "\n",
              "[10000 rows x 3 columns]>"
            ]
          },
          "metadata": {},
          "execution_count": 3
        }
      ],
      "source": [
        "data = pd.read_csv('/content/drive/MyDrive/376_project/labels.csv', dtype=str)\n",
        "data.id = data.id + '.tif'\n",
        "print(data.shape)\n",
        "data.head\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "p_xLY4CemaEm",
        "outputId": "3213a71c-6c1b-4f95-9402-85ad381880b6"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Total number of examples in training set : 8000\n",
            "Total number of examples in validation set : 1000\n",
            "Total number of examples in test set : 1000\n"
          ]
        }
      ],
      "source": [
        "train, val_test = train_test_split(data, train_size=0.8, random_state=376, shuffle=True)\n",
        "val, test = train_test_split(val_test, train_size=0.5, random_state=376, shuffle=True)\n",
        "\n",
        "print(f\"Total number of examples in training set : {train.shape[0]}\")\n",
        "print(f\"Total number of examples in validation set : {val.shape[0]}\")\n",
        "print(f\"Total number of examples in test set : {test.shape[0]}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "Sq64rVarmh4p"
      },
      "outputs": [],
      "source": [
        "train_datagen = ImageDataGenerator(rescale=1/255,\n",
        "                                   rotation_range=45,\n",
        "                                   vertical_flip=True,\n",
        "                                   horizontal_flip=True,\n",
        "                                   width_shift_range=0.3,\n",
        "                                   height_shift_range=0.3,\n",
        "                                   dtype=tf.float32)\n",
        "val_datagen = ImageDataGenerator(rescale=1/255,\n",
        "                                 dtype=tf.float32)\n",
        "test_datagen = ImageDataGenerator(rescale=1/255,\n",
        "                                  dtype=tf.float32)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "br_jJe2mqZfX",
        "outputId": "724392b0-9d4b-46b4-a04e-a412ecc345d0"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 8000 validated image filenames belonging to 2 classes.\n",
            "Found 1000 validated image filenames belonging to 2 classes.\n",
            "Found 1000 validated image filenames belonging to 2 classes.\n"
          ]
        }
      ],
      "source": [
        "train_loader = train_datagen.flow_from_dataframe(dataframe=train,\n",
        "                                                 directory=oldpath,\n",
        "                                                 x_col='id',\n",
        "                                                 y_col='label',\n",
        "                                                 batch_size=BATCH_SIZE,\n",
        "                                                 seed=376,\n",
        "                                                 shuffle=True,\n",
        "                                                 class_mode='categorical',\n",
        "                                                 target_size=(96,96))\n",
        "\n",
        "val_loader = val_datagen.flow_from_dataframe(dataframe=val,\n",
        "                                             directory=oldpath,\n",
        "                                             x_col='id',\n",
        "                                             y_col='label',\n",
        "                                             batch_size=BATCH_SIZE,\n",
        "                                             seed=376,\n",
        "                                             shuffle=True,\n",
        "                                             class_mode='categorical',\n",
        "                                             target_size=(96,96))\n",
        "\n",
        "test_loader = test_datagen.flow_from_dataframe(dataframe=test,\n",
        "                                               directory=oldpath,\n",
        "                                               x_col='id',\n",
        "                                               y_col='label',\n",
        "                                               batch_size=BATCH_SIZE,\n",
        "                                               seed=376,\n",
        "                                               shuffle=False,\n",
        "                                               class_mode='categorical',\n",
        "                                               target_size=(96,96))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mYIrdJqUEqUS",
        "outputId": "e0b72acc-a008-4c8f-c3db-851503ffd57b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg19/vgg19_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "80142336/80134624 [==============================] - 0s 0us/step\n",
            "80150528/80134624 [==============================] - 0s 0us/step\n"
          ]
        }
      ],
      "source": [
        "base_model = tf.keras.applications.vgg19.VGG19(\n",
        "                                          input_shape=(32,32,3),\n",
        "                                          include_top=False,\n",
        "                                          weights='imagenet'\n",
        ")\n",
        "base_model.trainable = False"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Z1U9j43sFraR",
        "outputId": "b19ae20e-9be3-40d2-9469-1383b34c2bb7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"vgg19\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 32, 32, 3)]       0         \n",
            "                                                                 \n",
            " block1_conv1 (Conv2D)       (None, 32, 32, 64)        1792      \n",
            "                                                                 \n",
            " block1_conv2 (Conv2D)       (None, 32, 32, 64)        36928     \n",
            "                                                                 \n",
            " block1_pool (MaxPooling2D)  (None, 16, 16, 64)        0         \n",
            "                                                                 \n",
            " block2_conv1 (Conv2D)       (None, 16, 16, 128)       73856     \n",
            "                                                                 \n",
            " block2_conv2 (Conv2D)       (None, 16, 16, 128)       147584    \n",
            "                                                                 \n",
            " block2_pool (MaxPooling2D)  (None, 8, 8, 128)         0         \n",
            "                                                                 \n",
            " block3_conv1 (Conv2D)       (None, 8, 8, 256)         295168    \n",
            "                                                                 \n",
            " block3_conv2 (Conv2D)       (None, 8, 8, 256)         590080    \n",
            "                                                                 \n",
            " block3_conv3 (Conv2D)       (None, 8, 8, 256)         590080    \n",
            "                                                                 \n",
            " block3_conv4 (Conv2D)       (None, 8, 8, 256)         590080    \n",
            "                                                                 \n",
            " block3_pool (MaxPooling2D)  (None, 4, 4, 256)         0         \n",
            "                                                                 \n",
            " block4_conv1 (Conv2D)       (None, 4, 4, 512)         1180160   \n",
            "                                                                 \n",
            " block4_conv2 (Conv2D)       (None, 4, 4, 512)         2359808   \n",
            "                                                                 \n",
            " block4_conv3 (Conv2D)       (None, 4, 4, 512)         2359808   \n",
            "                                                                 \n",
            " block4_conv4 (Conv2D)       (None, 4, 4, 512)         2359808   \n",
            "                                                                 \n",
            " block4_pool (MaxPooling2D)  (None, 2, 2, 512)         0         \n",
            "                                                                 \n",
            " block5_conv1 (Conv2D)       (None, 2, 2, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv2 (Conv2D)       (None, 2, 2, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv3 (Conv2D)       (None, 2, 2, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv4 (Conv2D)       (None, 2, 2, 512)         2359808   \n",
            "                                                                 \n",
            " block5_pool (MaxPooling2D)  (None, 1, 1, 512)         0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 20,024,384\n",
            "Trainable params: 0\n",
            "Non-trainable params: 20,024,384\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "base_model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "MYEdnVLjJQRL"
      },
      "outputs": [],
      "source": [
        "model = tf.keras.Sequential([\n",
        "    tf.keras.layers.CenterCrop(32,32),\n",
        "    base_model,\n",
        "    tf.keras.layers.Flatten(),\n",
        "    tf.keras.layers.Dense(8, activation='relu'),\n",
        "    tf.keras.layers.Dropout(0.5),\n",
        "    tf.keras.layers.Dense(4, activation='relu'),\n",
        "    tf.keras.layers.Dropout(0.5),\n",
        "    tf.keras.layers.BatchNormalization(),\n",
        "    tf.keras.layers.Dense(2,activation='softmax')\n",
        "])\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "lr = 1e-4\n",
        "model.compile(\n",
        "    optimizer = tf.keras.optimizers.Adam(learning_rate=lr),\n",
        "    loss ='categorical_crossentropy',\n",
        "    metrics = ['accuracy', tf.keras.metrics.AUC()],\n",
        ")\n",
        "model.build(input_shape = (None,96,96,3))\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1Udd-AyMyFEC",
        "outputId": "346f5a06-294c-48ce-851a-be2aa6bb79ca"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " center_crop (CenterCrop)    (None, 32, 32, 3)         0         \n",
            "                                                                 \n",
            " vgg19 (Functional)          (None, 1, 1, 512)         20024384  \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 512)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 8)                 4104      \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 8)                 0         \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 4)                 36        \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 4)                 0         \n",
            "                                                                 \n",
            " batch_normalization (BatchN  (None, 4)                16        \n",
            " ormalization)                                                   \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 2)                 10        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 20,028,550\n",
            "Trainable params: 4,158\n",
            "Non-trainable params: 20,024,392\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mO8PERrFQwJD",
        "outputId": "ee0c1a5d-9061-4443-8ad0-c6381b9eef29"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "125/125 [==============================] - 2703s 22s/step - loss: 0.9414 - accuracy: 0.5128 - auc: 0.5346 - val_loss: 0.6171 - val_accuracy: 0.7160 - val_auc: 0.7852\n",
            "Epoch 2/50\n",
            "125/125 [==============================] - 32s 258ms/step - loss: 0.8272 - accuracy: 0.5421 - auc: 0.5679 - val_loss: 0.6401 - val_accuracy: 0.6180 - val_auc: 0.6845\n",
            "Epoch 3/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.7554 - accuracy: 0.5673 - auc: 0.6105 - val_loss: 0.6322 - val_accuracy: 0.6340 - val_auc: 0.6987\n",
            "Epoch 4/50\n",
            "125/125 [==============================] - 32s 253ms/step - loss: 0.7204 - accuracy: 0.5692 - auc: 0.6183 - val_loss: 0.6015 - val_accuracy: 0.6830 - val_auc: 0.7521\n",
            "Epoch 5/50\n",
            "125/125 [==============================] - 32s 252ms/step - loss: 0.6984 - accuracy: 0.5785 - auc: 0.6281 - val_loss: 0.5910 - val_accuracy: 0.6890 - val_auc: 0.7675\n",
            "Epoch 6/50\n",
            "125/125 [==============================] - 32s 252ms/step - loss: 0.6872 - accuracy: 0.5867 - auc: 0.6448 - val_loss: 0.5776 - val_accuracy: 0.7030 - val_auc: 0.7923\n",
            "Epoch 7/50\n",
            "125/125 [==============================] - 32s 260ms/step - loss: 0.6791 - accuracy: 0.5864 - auc: 0.6388 - val_loss: 0.5726 - val_accuracy: 0.7130 - val_auc: 0.7997\n",
            "Epoch 8/50\n",
            "125/125 [==============================] - 32s 252ms/step - loss: 0.6763 - accuracy: 0.5901 - auc: 0.6523 - val_loss: 0.5688 - val_accuracy: 0.7190 - val_auc: 0.8052\n",
            "Epoch 9/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.6616 - accuracy: 0.5895 - auc: 0.6477 - val_loss: 0.5725 - val_accuracy: 0.7220 - val_auc: 0.8055\n",
            "Epoch 10/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.6556 - accuracy: 0.6036 - auc: 0.6647 - val_loss: 0.5673 - val_accuracy: 0.7260 - val_auc: 0.8123\n",
            "Epoch 11/50\n",
            "125/125 [==============================] - 32s 253ms/step - loss: 0.6578 - accuracy: 0.6070 - auc: 0.6703 - val_loss: 0.5627 - val_accuracy: 0.7400 - val_auc: 0.8114\n",
            "Epoch 12/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.6522 - accuracy: 0.6114 - auc: 0.6751 - val_loss: 0.5664 - val_accuracy: 0.7390 - val_auc: 0.8115\n",
            "Epoch 13/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6528 - accuracy: 0.6223 - auc: 0.6814 - val_loss: 0.5614 - val_accuracy: 0.7380 - val_auc: 0.8054\n",
            "Epoch 14/50\n",
            "125/125 [==============================] - 32s 252ms/step - loss: 0.6478 - accuracy: 0.6212 - auc: 0.6854 - val_loss: 0.5633 - val_accuracy: 0.7370 - val_auc: 0.8074\n",
            "Epoch 15/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6435 - accuracy: 0.6204 - auc: 0.6873 - val_loss: 0.5688 - val_accuracy: 0.7420 - val_auc: 0.8029\n",
            "Epoch 16/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6485 - accuracy: 0.6106 - auc: 0.6781 - val_loss: 0.5643 - val_accuracy: 0.7430 - val_auc: 0.8008\n",
            "Epoch 17/50\n",
            "125/125 [==============================] - 32s 252ms/step - loss: 0.6478 - accuracy: 0.6246 - auc: 0.6886 - val_loss: 0.5619 - val_accuracy: 0.7420 - val_auc: 0.7989\n",
            "Epoch 18/50\n",
            "125/125 [==============================] - 31s 249ms/step - loss: 0.6436 - accuracy: 0.6154 - auc: 0.6841 - val_loss: 0.5653 - val_accuracy: 0.7430 - val_auc: 0.8004\n",
            "Epoch 19/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6458 - accuracy: 0.6066 - auc: 0.6755 - val_loss: 0.5676 - val_accuracy: 0.7450 - val_auc: 0.7987\n",
            "Epoch 20/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6392 - accuracy: 0.6087 - auc: 0.6856 - val_loss: 0.5655 - val_accuracy: 0.7420 - val_auc: 0.7956\n",
            "Epoch 21/50\n",
            "125/125 [==============================] - 32s 256ms/step - loss: 0.6359 - accuracy: 0.6115 - auc: 0.6825 - val_loss: 0.5658 - val_accuracy: 0.7340 - val_auc: 0.7875\n",
            "Epoch 22/50\n",
            "125/125 [==============================] - 32s 258ms/step - loss: 0.6426 - accuracy: 0.6135 - auc: 0.6828 - val_loss: 0.5648 - val_accuracy: 0.7370 - val_auc: 0.7964\n",
            "Epoch 23/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6298 - accuracy: 0.6216 - auc: 0.6936 - val_loss: 0.5640 - val_accuracy: 0.7350 - val_auc: 0.7894\n",
            "Epoch 24/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6388 - accuracy: 0.6155 - auc: 0.6855 - val_loss: 0.5671 - val_accuracy: 0.7390 - val_auc: 0.7961\n",
            "Epoch 25/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6287 - accuracy: 0.6127 - auc: 0.6874 - val_loss: 0.5640 - val_accuracy: 0.7360 - val_auc: 0.7900\n",
            "Epoch 26/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6400 - accuracy: 0.6191 - auc: 0.6893 - val_loss: 0.5652 - val_accuracy: 0.7430 - val_auc: 0.7950\n",
            "Epoch 27/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6367 - accuracy: 0.6174 - auc: 0.6886 - val_loss: 0.5659 - val_accuracy: 0.7440 - val_auc: 0.7917\n",
            "Epoch 28/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6324 - accuracy: 0.6233 - auc: 0.6956 - val_loss: 0.5643 - val_accuracy: 0.7390 - val_auc: 0.7905\n",
            "Epoch 29/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6332 - accuracy: 0.6150 - auc: 0.6893 - val_loss: 0.5667 - val_accuracy: 0.7380 - val_auc: 0.7911\n",
            "Epoch 30/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.6261 - accuracy: 0.6237 - auc: 0.6974 - val_loss: 0.5661 - val_accuracy: 0.7420 - val_auc: 0.7938\n",
            "Epoch 31/50\n",
            "125/125 [==============================] - 33s 262ms/step - loss: 0.6288 - accuracy: 0.6212 - auc: 0.7001 - val_loss: 0.5647 - val_accuracy: 0.7470 - val_auc: 0.7987\n",
            "Epoch 32/50\n",
            "125/125 [==============================] - 32s 253ms/step - loss: 0.6259 - accuracy: 0.6260 - auc: 0.7023 - val_loss: 0.5659 - val_accuracy: 0.7470 - val_auc: 0.7980\n",
            "Epoch 33/50\n",
            "125/125 [==============================] - 31s 249ms/step - loss: 0.6306 - accuracy: 0.6332 - auc: 0.7005 - val_loss: 0.5639 - val_accuracy: 0.7330 - val_auc: 0.7922\n",
            "Epoch 34/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6240 - accuracy: 0.6241 - auc: 0.6993 - val_loss: 0.5658 - val_accuracy: 0.7500 - val_auc: 0.8025\n",
            "Epoch 35/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6363 - accuracy: 0.6173 - auc: 0.6889 - val_loss: 0.5673 - val_accuracy: 0.7540 - val_auc: 0.8022\n",
            "Epoch 36/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6306 - accuracy: 0.6105 - auc: 0.6826 - val_loss: 0.5650 - val_accuracy: 0.7360 - val_auc: 0.7914\n",
            "Epoch 37/50\n",
            "125/125 [==============================] - 31s 250ms/step - loss: 0.6278 - accuracy: 0.6190 - auc: 0.6941 - val_loss: 0.5663 - val_accuracy: 0.7490 - val_auc: 0.7973\n",
            "Epoch 38/50\n",
            "125/125 [==============================] - 32s 256ms/step - loss: 0.6294 - accuracy: 0.6229 - auc: 0.6972 - val_loss: 0.5643 - val_accuracy: 0.7460 - val_auc: 0.7931\n",
            "Epoch 39/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6319 - accuracy: 0.6099 - auc: 0.6884 - val_loss: 0.5673 - val_accuracy: 0.7470 - val_auc: 0.8008\n",
            "Epoch 40/50\n",
            "125/125 [==============================] - 32s 255ms/step - loss: 0.6301 - accuracy: 0.6301 - auc: 0.6982 - val_loss: 0.5682 - val_accuracy: 0.7480 - val_auc: 0.7940\n",
            "Epoch 41/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6238 - accuracy: 0.6291 - auc: 0.7016 - val_loss: 0.5681 - val_accuracy: 0.7470 - val_auc: 0.8029\n",
            "Epoch 42/50\n",
            "125/125 [==============================] - 31s 252ms/step - loss: 0.6239 - accuracy: 0.6121 - auc: 0.6889 - val_loss: 0.5653 - val_accuracy: 0.7350 - val_auc: 0.7956\n",
            "Epoch 43/50\n",
            "125/125 [==============================] - 31s 249ms/step - loss: 0.6331 - accuracy: 0.6244 - auc: 0.6968 - val_loss: 0.5635 - val_accuracy: 0.7350 - val_auc: 0.7945\n",
            "Epoch 44/50\n",
            "125/125 [==============================] - 31s 249ms/step - loss: 0.6252 - accuracy: 0.6311 - auc: 0.7036 - val_loss: 0.5636 - val_accuracy: 0.7490 - val_auc: 0.8045\n",
            "Epoch 45/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6256 - accuracy: 0.6352 - auc: 0.7073 - val_loss: 0.5643 - val_accuracy: 0.7470 - val_auc: 0.8015\n",
            "Epoch 46/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6237 - accuracy: 0.6267 - auc: 0.7004 - val_loss: 0.5651 - val_accuracy: 0.7530 - val_auc: 0.7995\n",
            "Epoch 47/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6214 - accuracy: 0.6201 - auc: 0.6975 - val_loss: 0.5681 - val_accuracy: 0.7490 - val_auc: 0.8002\n",
            "Epoch 48/50\n",
            "125/125 [==============================] - 32s 254ms/step - loss: 0.6223 - accuracy: 0.6236 - auc: 0.6991 - val_loss: 0.5625 - val_accuracy: 0.7480 - val_auc: 0.7988\n",
            "Epoch 49/50\n",
            "125/125 [==============================] - 31s 251ms/step - loss: 0.6195 - accuracy: 0.6275 - auc: 0.7020 - val_loss: 0.5647 - val_accuracy: 0.7650 - val_auc: 0.8091\n",
            "Epoch 50/50\n",
            "125/125 [==============================] - 32s 256ms/step - loss: 0.6224 - accuracy: 0.6274 - auc: 0.7043 - val_loss: 0.5645 - val_accuracy: 0.7530 - val_auc: 0.8040\n"
          ]
        }
      ],
      "source": [
        "TR_STEPS = len(train_loader)\n",
        "VA_STEPS = len(val_loader)\n",
        "epochs = 50\n",
        "history = model.fit(\n",
        "    x = train_loader,\n",
        "    steps_per_epoch = TR_STEPS,\n",
        "    validation_data = val_loader,\n",
        "    validation_steps = VA_STEPS,\n",
        "    verbose = 1,\n",
        "    epochs=epochs\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.evaluate(test_loader)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iIFK6MF0D2CU",
        "outputId": "5e198284-8da4-4485-d957-ddba386077f0"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "16/16 [==============================] - 294s 20s/step - loss: 0.5723 - accuracy: 0.7410 - auc: 0.7876\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.572319507598877, 0.7409999966621399, 0.7876184582710266]"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "history_dict = history.history\n",
        "print(history_dict.keys())\n",
        "\n",
        "acc = history_dict['accuracy']\n",
        "val_acc = history_dict['val_accuracy']\n",
        "loss = history_dict['loss']\n",
        "val_loss = history_dict['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "fig = plt.figure(figsize=(10, 6))\n",
        "fig.tight_layout()\n",
        "\n",
        "plt.subplot(2, 1, 1)\n",
        "# r is for \"solid red line\"\n",
        "plt.plot(epochs, loss, 'r', label='Training loss')\n",
        "# b is for \"solid blue line\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
        "plt.title('Training and validation loss')\n",
        "# plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.subplot(2, 1, 2)\n",
        "plt.plot(epochs, acc, 'r', label='Training acc')\n",
        "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "id": "EFHwUL_EP9Kl",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 439
        },
        "outputId": "1b73de8b-9e33-4da2-fef1-f675038c59e8"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy', 'auc', 'val_loss', 'val_accuracy', 'val_auc'])\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fa8daf6c310>"
            ]
          },
          "metadata": {},
          "execution_count": 13
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x432 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.save('/content/drive/MyDrive/376_project/vgg19_crop/')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "awoYtqQrAKad",
        "outputId": "e4e783e6-11f8-403b-93d6-4682bc36f35c"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "INFO:tensorflow:Assets written to: /content/drive/MyDrive/376_project/vgg19_no_crop/assets\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model = tf.keras.models.load_model('/content/drive/MyDrive/376_project/vgg19_crop/')\n",
        "model.evaluate(test_loader)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JtB-KgISATw5",
        "outputId": "3ffa71cc-b938-4e0b-a341-4ee0c81db440"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "16/16 [==============================] - 2s 108ms/step - loss: 0.5723 - accuracy: 0.7410 - auc: 0.7876\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.572319507598877, 0.7409999966621399, 0.7876184582710266]"
            ]
          },
          "metadata": {},
          "execution_count": 16
        }
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": " VGG19_cancer_crop.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}