3_resnet50(colab).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## This Notebook Uses ReSNet50 (Transfer Learning with PreTrained Weights and Retraining) Model to generate Output Predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  <font color='red'>Table of Contents</font>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "I3BHeMMtWiIE"
   },
   "source": [
    "[9. ResNet50 Models](#section9)<br>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 104
    },
    "colab_type": "code",
    "id": "22ccXpX_WQ54",
    "outputId": "223b61e2-9b29-4677-8683-c8ad021ba2f0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2  CPU cores\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
      "  import pandas.util.testing as tm\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Research Kernel Link - https://github.com/dimitreOliveira/APTOS2019BlindnessDetection/blob/master/Model%20backlog/ResNet50/4%20-%20ResNet50%20-%20Batch%20size%2022.ipynb\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "from prettytable import PrettyTable\n",
    "import pickle\n",
    "import multiprocessing\n",
    "from multiprocessing.pool import ThreadPool\n",
    "print(multiprocessing.cpu_count(),\" CPU cores\")\n",
    "\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.rcParams[\"axes.grid\"] = False\n",
    "\n",
    "from sklearn.metrics import confusion_matrix, cohen_kappa_score,accuracy_score\n",
    "\n",
    "from PIL import Image\n",
    "import cv2\n",
    "\n",
    "import keras\n",
    "from keras import applications\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras import optimizers,Model,Sequential\n",
    "from keras.layers import Input,GlobalAveragePooling2D,Dropout,Dense,Activation\n",
    "from keras.callbacks.callbacks import EarlyStopping,ReduceLROnPlateau\n",
    "\n",
    "# Colab Libs...\n",
    "from pydrive.auth import GoogleAuth\n",
    "from pydrive.drive import GoogleDrive\n",
    "from google.colab import auth\n",
    "from oauth2client.client import GoogleCredentials"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "yRu-rH5tWQ58"
   },
   "source": [
    "#  <a id = 'section9'> <font color='red'>  9. ResNet50 Models  </font> </a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ej-CCMUPW__D"
   },
   "source": [
    "### <font color='red'> 9.1 Setup Colab Environment </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "AOPuWlg2W43_"
   },
   "outputs": [],
   "source": [
    "# Importing Libraries\n",
    "#ref - https://buomsoo-kim.github.io/colab/2018/04/16/Importing-files-from-Google-Drive-in-Google-Colab.md/\n",
    "\n",
    "auth.authenticate_user()\n",
    "gauth = GoogleAuth()\n",
    "gauth.credentials = GoogleCredentials.get_application_default()\n",
    "drive = GoogleDrive(gauth)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 121
    },
    "colab_type": "code",
    "id": "P5s9LrQAXEBl",
    "outputId": "1512d185-9d51-4dea-f89b-fc9ced7ad2cb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly\n",
      "\n",
      "Enter your authorization code:\n",
      "··········\n",
      "Mounted at /content/gdrive\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/gdrive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "UnaOi7ApXEEn",
    "outputId": "b3809493-3a98-42d8-d647-33745d9c181d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We are currently in the folder of  /content/gdrive/My Drive/aptos2019\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "os.chdir('/content/gdrive/My Drive/aptos2019')\n",
    "print(\"We are currently in the folder of \",os.getcwd())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0cybxJXNWQ5-"
   },
   "outputs": [],
   "source": [
    "def load_data():\n",
    "    file = open('df_train_train', 'rb')\n",
    "    df_train_train = pickle.load(file)\n",
    "    file.close()\n",
    "\n",
    "    file = open('df_train_test', 'rb')\n",
    "    df_train_test = pickle.load(file)\n",
    "    file.close()\n",
    "    \n",
    "    return df_train_train,df_train_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 258
    },
    "colab_type": "code",
    "id": "z1F6Uye-WQ6D",
    "outputId": "ed9b014f-f77d-4ceb-a114-7f4e7b8e20f0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2929, 4) (733, 4) \n",
      "\n"
     ]
    },
    {
     "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>id_code</th>\n",
       "      <th>diagnosis</th>\n",
       "      <th>file_path</th>\n",
       "      <th>file_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3396</th>\n",
       "      <td>eb6b1f1c09db</td>\n",
       "      <td>0</td>\n",
       "      <td>./train_images/eb6b1f1c09db.png</td>\n",
       "      <td>eb6b1f1c09db.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>07a3be30563b</td>\n",
       "      <td>2</td>\n",
       "      <td>./train_images/07a3be30563b.png</td>\n",
       "      <td>07a3be30563b.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3204</th>\n",
       "      <td>df3adfd6ba36</td>\n",
       "      <td>2</td>\n",
       "      <td>./train_images/df3adfd6ba36.png</td>\n",
       "      <td>df3adfd6ba36.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3324</th>\n",
       "      <td>e68bdd36e589</td>\n",
       "      <td>0</td>\n",
       "      <td>./train_images/e68bdd36e589.png</td>\n",
       "      <td>e68bdd36e589.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1911</th>\n",
       "      <td>86e7f98f73f1</td>\n",
       "      <td>2</td>\n",
       "      <td>./train_images/86e7f98f73f1.png</td>\n",
       "      <td>86e7f98f73f1.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>396</th>\n",
       "      <td>1caba2fb38f6</td>\n",
       "      <td>0</td>\n",
       "      <td>./train_images/1caba2fb38f6.png</td>\n",
       "      <td>1caba2fb38f6.png</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           id_code diagnosis                        file_path         file_name\n",
       "3396  eb6b1f1c09db         0  ./train_images/eb6b1f1c09db.png  eb6b1f1c09db.png\n",
       "100   07a3be30563b         2  ./train_images/07a3be30563b.png  07a3be30563b.png\n",
       "3204  df3adfd6ba36         2  ./train_images/df3adfd6ba36.png  df3adfd6ba36.png\n",
       "3324  e68bdd36e589         0  ./train_images/e68bdd36e589.png  e68bdd36e589.png\n",
       "1911  86e7f98f73f1         2  ./train_images/86e7f98f73f1.png  86e7f98f73f1.png\n",
       "396   1caba2fb38f6         0  ./train_images/1caba2fb38f6.png  1caba2fb38f6.png"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train_train,df_train_test = load_data()\n",
    "print(df_train_train.shape,df_train_test.shape,'\\n')\n",
    "df_train_train.head(6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3H9xIXaaMmza"
   },
   "source": [
    "### <font color='red'> 9.2 Image Pre Processing </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "SAct4MuUNjH4"
   },
   "outputs": [],
   "source": [
    "IMG_SIZE = 512"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "L6lBJR9tMo68"
   },
   "outputs": [],
   "source": [
    "def crop_image_from_gray(img,tol=7):\n",
    "    if img.ndim ==2:\n",
    "        mask = img>tol\n",
    "        return img[np.ix_(mask.any(1),mask.any(0))]\n",
    "    elif img.ndim==3:\n",
    "        gray_img = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)\n",
    "        mask = gray_img>tol\n",
    "        \n",
    "        check_shape = img[:,:,0][np.ix_(mask.any(1),mask.any(0))].shape[0]\n",
    "        if (check_shape == 0): # image is too dark so that we crop out everything,\n",
    "            return img # return original image\n",
    "        else:\n",
    "            img1=img[:,:,0][np.ix_(mask.any(1),mask.any(0))]\n",
    "            img2=img[:,:,1][np.ix_(mask.any(1),mask.any(0))]\n",
    "            img3=img[:,:,2][np.ix_(mask.any(1),mask.any(0))]\n",
    "    #         print(img1.shape,img2.shape,img3.shape)\n",
    "            img = np.stack([img1,img2,img3],axis=-1)\n",
    "    #         print(img.shape)\n",
    "        return img\n",
    "\n",
    "def circle_crop(img, sigmaX = 30):   \n",
    "    \"\"\"\n",
    "    Create circular crop around image centre    \n",
    "    \"\"\"    \n",
    "    img = crop_image_from_gray(img)    \n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    \n",
    "    height, width, depth = img.shape    \n",
    "    \n",
    "    x = int(width/2)\n",
    "    y = int(height/2)\n",
    "    r = np.amin((x,y))\n",
    "    \n",
    "    circle_img = np.zeros((height, width), np.uint8)\n",
    "    cv2.circle(circle_img, (x,y), int(r), 1, thickness=-1)\n",
    "    img = cv2.bitwise_and(img, img, mask=circle_img)\n",
    "    img = crop_image_from_gray(img)\n",
    "    img=cv2.addWeighted(img,4, cv2.GaussianBlur( img , (0,0) , sigmaX) ,-4 ,128)\n",
    "    return img \n",
    "\n",
    "def preprocess_image(file):\n",
    "    input_filepath = os.path.join('./','train_images_resized','{}.png'.format(file))\n",
    "    output_filepath = os.path.join('./','train_images_resized_preprocessed','{}.png'.format(file))\n",
    "    \n",
    "    img = cv2.imread(input_filepath)\n",
    "    img = circle_crop(img) \n",
    "    cv2.imwrite(output_filepath, cv2.resize(img, (IMG_SIZE,IMG_SIZE)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-VIgIRbCNBix"
   },
   "outputs": [],
   "source": [
    "'''This Function uses Multi processing for faster saving of images into folder'''\n",
    "\n",
    "def multiprocess_image_processor(process:int, imgs:list):\n",
    "    \"\"\"\n",
    "    Inputs:\n",
    "        process: (int) number of process to run\n",
    "        imgs:(list) list of images\n",
    "    \"\"\"\n",
    "    print(f'MESSAGE: Running {process} process')\n",
    "    results = ThreadPool(process).map(preprocess_image, imgs)\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Cf_-VV_DNC0N"
   },
   "outputs": [],
   "source": [
    "# Use 2 cores (colab)\n",
    "# ref - https://stackoverflow.com/questions/1006289/how-to-find-out-the-number-of-cpus-using-python\n",
    "\n",
    "#multiprocess_image_processor(2, list(df_train_train.id_code.values))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Oq5liZxiWQ6H"
   },
   "source": [
    "### <font color='red'> 9.3 Train Model </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "KBX1kOjkWQ6H"
   },
   "outputs": [],
   "source": [
    "# Model parameters\n",
    "BATCH_SIZE = 8\n",
    "EPOCHS = 40\n",
    "WARMUP_EPOCHS = 2\n",
    "LEARNING_RATE = 1e-4\n",
    "WARMUP_LEARNING_RATE = 1e-3\n",
    "HEIGHT = 320\n",
    "WIDTH = 320\n",
    "CANAL = 3\n",
    "N_CLASSES = df_train_train['diagnosis'].nunique()\n",
    "ES_PATIENCE = 5\n",
    "RLROP_PATIENCE = 3\n",
    "DECAY_DROP = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "brHRA_U0WQ6K"
   },
   "outputs": [],
   "source": [
    "def img_generator(train,test):\n",
    "    train_datagen=ImageDataGenerator(rescale=1./255, validation_split=0.2,horizontal_flip=True)\n",
    "    \n",
    "    train_generator=train_datagen.flow_from_dataframe(dataframe=df_train_train,\n",
    "                                                      directory=\"./train_images_resized_preprocessed/\",\n",
    "                                                      x_col=\"file_name\",\n",
    "                                                      y_col=\"diagnosis\",\n",
    "                                                      batch_size=BATCH_SIZE,\n",
    "                                                      class_mode=\"categorical\",\n",
    "                                                      target_size=(HEIGHT, WIDTH),\n",
    "                                                      subset='training')\n",
    "    \n",
    "    valid_generator=train_datagen.flow_from_dataframe(dataframe=df_train_train,\n",
    "                                                      directory=\"./train_images_resized_preprocessed/\",\n",
    "                                                      x_col=\"file_name\",\n",
    "                                                      y_col=\"diagnosis\",\n",
    "                                                      batch_size=BATCH_SIZE,\n",
    "                                                      class_mode=\"categorical\",    \n",
    "                                                      target_size=(HEIGHT, WIDTH),\n",
    "                                                      subset='validation')\n",
    "    \n",
    "    test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "    test_generator = test_datagen.flow_from_dataframe(dataframe=df_train_test,\n",
    "                                                      directory = \"./test_images_resized_preprocessed/\",\n",
    "                                                      x_col=\"file_name\",\n",
    "                                                      target_size=(HEIGHT, WIDTH),\n",
    "                                                      batch_size=1,\n",
    "                                                      shuffle=False,\n",
    "                                                      class_mode=None)\n",
    "    \n",
    "    return train_generator,valid_generator,test_generator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 67
    },
    "colab_type": "code",
    "id": "CQjnOOEhWQ6P",
    "outputId": "703a18c6-c2d3-4d5a-c59e-26bc0d684180"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2344 validated image filenames belonging to 5 classes.\n",
      "Found 585 validated image filenames belonging to 5 classes.\n",
      "Found 733 validated image filenames.\n"
     ]
    }
   ],
   "source": [
    "train_generator,valid_generator,test_generator = img_generator(df_train_train,df_train_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "p1ktmE4ZWQ6T"
   },
   "outputs": [],
   "source": [
    "def create_model(input_shape, n_out):\n",
    "    input_tensor = Input(shape=input_shape)\n",
    "    base_model = applications.ResNet50(weights=None, include_top=False,input_tensor=input_tensor)\n",
    "    base_model.load_weights('resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5')\n",
    "\n",
    "    x = GlobalAveragePooling2D()(base_model.output)\n",
    "    x = Dropout(0.5)(x)\n",
    "    x = Dense(2048, activation='relu')(x)\n",
    "    x = Dropout(0.5)(x)\n",
    "    final_output = Dense(n_out, activation='softmax', name='final_output')(x)\n",
    "    model = Model(input_tensor, final_output)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "GIHj-I3LWQ6W",
    "outputId": "438f8279-a839-406e-92a0-b28a6c13f7c6"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/keras_applications/resnet50.py:265: UserWarning: The output shape of `ResNet50(include_top=False)` has been changed since Keras 2.2.0.\n",
      "  warnings.warn('The output shape of `ResNet50(include_top=False)` '\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 320, 320, 3)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1_pad (ZeroPadding2D)       (None, 326, 326, 3)  0           input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1 (Conv2D)                  (None, 160, 160, 64) 9472        conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "bn_conv1 (BatchNormalization)   (None, 160, 160, 64) 256         conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_50 (Activation)      (None, 160, 160, 64) 0           bn_conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "pool1_pad (ZeroPadding2D)       (None, 162, 162, 64) 0           activation_50[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2D)  (None, 80, 80, 64)   0           pool1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2a (Conv2D)         (None, 80, 80, 64)   4160        max_pooling2d_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_51 (Activation)      (None, 80, 80, 64)   0           bn2a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_51[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_52 (Activation)      (None, 80, 80, 64)   0           bn2a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_52[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch1 (Conv2D)          (None, 80, 80, 256)  16640       max_pooling2d_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch1 (BatchNormalizatio (None, 80, 80, 256)  1024        res2a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_17 (Add)                    (None, 80, 80, 256)  0           bn2a_branch2c[0][0]              \n",
      "                                                                 bn2a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_53 (Activation)      (None, 80, 80, 256)  0           add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2a (Conv2D)         (None, 80, 80, 64)   16448       activation_53[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_54 (Activation)      (None, 80, 80, 64)   0           bn2b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_54[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_55 (Activation)      (None, 80, 80, 64)   0           bn2b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_55[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_18 (Add)                    (None, 80, 80, 256)  0           bn2b_branch2c[0][0]              \n",
      "                                                                 activation_53[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_56 (Activation)      (None, 80, 80, 256)  0           add_18[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2a (Conv2D)         (None, 80, 80, 64)   16448       activation_56[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_57 (Activation)      (None, 80, 80, 64)   0           bn2c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_57[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_58 (Activation)      (None, 80, 80, 64)   0           bn2c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_58[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_19 (Add)                    (None, 80, 80, 256)  0           bn2c_branch2c[0][0]              \n",
      "                                                                 activation_56[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_59 (Activation)      (None, 80, 80, 256)  0           add_19[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2a (Conv2D)         (None, 40, 40, 128)  32896       activation_59[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_60 (Activation)      (None, 40, 40, 128)  0           bn3a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_60[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_61 (Activation)      (None, 40, 40, 128)  0           bn3a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_61[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch1 (Conv2D)          (None, 40, 40, 512)  131584      activation_59[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch1 (BatchNormalizatio (None, 40, 40, 512)  2048        res3a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_20 (Add)                    (None, 40, 40, 512)  0           bn3a_branch2c[0][0]              \n",
      "                                                                 bn3a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_62 (Activation)      (None, 40, 40, 512)  0           add_20[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_62[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_63 (Activation)      (None, 40, 40, 128)  0           bn3b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_63[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_64 (Activation)      (None, 40, 40, 128)  0           bn3b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_64[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_21 (Add)                    (None, 40, 40, 512)  0           bn3b_branch2c[0][0]              \n",
      "                                                                 activation_62[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_65 (Activation)      (None, 40, 40, 512)  0           add_21[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_65[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_66 (Activation)      (None, 40, 40, 128)  0           bn3c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_66[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_67 (Activation)      (None, 40, 40, 128)  0           bn3c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_67[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_22 (Add)                    (None, 40, 40, 512)  0           bn3c_branch2c[0][0]              \n",
      "                                                                 activation_65[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_68 (Activation)      (None, 40, 40, 512)  0           add_22[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_68[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_69 (Activation)      (None, 40, 40, 128)  0           bn3d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_69[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_70 (Activation)      (None, 40, 40, 128)  0           bn3d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_70[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_23 (Add)                    (None, 40, 40, 512)  0           bn3d_branch2c[0][0]              \n",
      "                                                                 activation_68[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_71 (Activation)      (None, 40, 40, 512)  0           add_23[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2a (Conv2D)         (None, 20, 20, 256)  131328      activation_71[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_72 (Activation)      (None, 20, 20, 256)  0           bn4a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_72[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_73 (Activation)      (None, 20, 20, 256)  0           bn4a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_73[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch1 (Conv2D)          (None, 20, 20, 1024) 525312      activation_71[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch1 (BatchNormalizatio (None, 20, 20, 1024) 4096        res4a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_24 (Add)                    (None, 20, 20, 1024) 0           bn4a_branch2c[0][0]              \n",
      "                                                                 bn4a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_74 (Activation)      (None, 20, 20, 1024) 0           add_24[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_74[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_75 (Activation)      (None, 20, 20, 256)  0           bn4b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_75[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_76 (Activation)      (None, 20, 20, 256)  0           bn4b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_76[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_25 (Add)                    (None, 20, 20, 1024) 0           bn4b_branch2c[0][0]              \n",
      "                                                                 activation_74[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_77 (Activation)      (None, 20, 20, 1024) 0           add_25[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_77[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_78 (Activation)      (None, 20, 20, 256)  0           bn4c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_78[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_79 (Activation)      (None, 20, 20, 256)  0           bn4c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_79[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_26 (Add)                    (None, 20, 20, 1024) 0           bn4c_branch2c[0][0]              \n",
      "                                                                 activation_77[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_80 (Activation)      (None, 20, 20, 1024) 0           add_26[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_80[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_81 (Activation)      (None, 20, 20, 256)  0           bn4d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_81[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_82 (Activation)      (None, 20, 20, 256)  0           bn4d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_82[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_27 (Add)                    (None, 20, 20, 1024) 0           bn4d_branch2c[0][0]              \n",
      "                                                                 activation_80[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_83 (Activation)      (None, 20, 20, 1024) 0           add_27[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_83[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4e_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_84 (Activation)      (None, 20, 20, 256)  0           bn4e_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_84[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4e_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_85 (Activation)      (None, 20, 20, 256)  0           bn4e_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_85[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4e_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_28 (Add)                    (None, 20, 20, 1024) 0           bn4e_branch2c[0][0]              \n",
      "                                                                 activation_83[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_86 (Activation)      (None, 20, 20, 1024) 0           add_28[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_86[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4f_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_87 (Activation)      (None, 20, 20, 256)  0           bn4f_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_87[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4f_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_88 (Activation)      (None, 20, 20, 256)  0           bn4f_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_88[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4f_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_29 (Add)                    (None, 20, 20, 1024) 0           bn4f_branch2c[0][0]              \n",
      "                                                                 activation_86[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_89 (Activation)      (None, 20, 20, 1024) 0           add_29[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2a (Conv2D)         (None, 10, 10, 512)  524800      activation_89[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_90 (Activation)      (None, 10, 10, 512)  0           bn5a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_90[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_91 (Activation)      (None, 10, 10, 512)  0           bn5a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_91[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch1 (Conv2D)          (None, 10, 10, 2048) 2099200     activation_89[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch1 (BatchNormalizatio (None, 10, 10, 2048) 8192        res5a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_30 (Add)                    (None, 10, 10, 2048) 0           bn5a_branch2c[0][0]              \n",
      "                                                                 bn5a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_92 (Activation)      (None, 10, 10, 2048) 0           add_30[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2a (Conv2D)         (None, 10, 10, 512)  1049088     activation_92[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_93 (Activation)      (None, 10, 10, 512)  0           bn5b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_93[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_94 (Activation)      (None, 10, 10, 512)  0           bn5b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_94[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_31 (Add)                    (None, 10, 10, 2048) 0           bn5b_branch2c[0][0]              \n",
      "                                                                 activation_92[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_95 (Activation)      (None, 10, 10, 2048) 0           add_31[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2a (Conv2D)         (None, 10, 10, 512)  1049088     activation_95[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_96 (Activation)      (None, 10, 10, 512)  0           bn5c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_96[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_97 (Activation)      (None, 10, 10, 512)  0           bn5c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_97[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_32 (Add)                    (None, 10, 10, 2048) 0           bn5c_branch2c[0][0]              \n",
      "                                                                 activation_95[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_98 (Activation)      (None, 10, 10, 2048) 0           add_32[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d_2 (Glo (None, 2048)         0           activation_98[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 2048)         0           global_average_pooling2d_2[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 2048)         4196352     dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 2048)         0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "final_output (Dense)            (None, 5)            10245       dropout_4[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 27,794,309\n",
      "Trainable params: 4,206,597\n",
      "Non-trainable params: 23,587,712\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = create_model(input_shape=(HEIGHT, WIDTH, CANAL), n_out=N_CLASSES)\n",
    "\n",
    "for layer in model.layers:\n",
    "    layer.trainable = False\n",
    "\n",
    "for i in range(-5, 0):\n",
    "    model.layers[i].trainable = True\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "FdOZMTWQWQ6Y",
    "outputId": "669c9c51-fa6a-463b-d9e8-770aea573a0b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "293 73\n"
     ]
    }
   ],
   "source": [
    "STEP_SIZE_TRAIN = train_generator.n//train_generator.batch_size\n",
    "STEP_SIZE_VALID = valid_generator.n//valid_generator.batch_size\n",
    "print(STEP_SIZE_TRAIN,STEP_SIZE_VALID)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 84
    },
    "colab_type": "code",
    "id": "UR-WhqwSWQ6a",
    "outputId": "3644a7a6-41c8-4449-9674-30613a0d9b1e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/2\n",
      "293/293 [==============================] - 1422s 5s/step - loss: 1.4534 - accuracy: 0.5943 - val_loss: 1.6658 - val_accuracy: 0.4897\n",
      "Epoch 2/2\n",
      "293/293 [==============================] - 68s 233ms/step - loss: 0.8575 - accuracy: 0.6788 - val_loss: 1.7975 - val_accuracy: 0.4905\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer = optimizers.Adam(lr=WARMUP_LEARNING_RATE),loss = 'categorical_crossentropy',metrics = ['accuracy'])\n",
    "\n",
    "history_warmup = model.fit_generator(generator=train_generator,\n",
    "                                     steps_per_epoch=STEP_SIZE_TRAIN,\n",
    "                                     validation_data=valid_generator,validation_steps=STEP_SIZE_VALID,\n",
    "                                     epochs=WARMUP_EPOCHS,\n",
    "                                     verbose=1).history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "I0EwusGsWQ6c",
    "outputId": "265ae8eb-738f-48d1-93a8-9db67513a68d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 320, 320, 3)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1_pad (ZeroPadding2D)       (None, 326, 326, 3)  0           input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1 (Conv2D)                  (None, 160, 160, 64) 9472        conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "bn_conv1 (BatchNormalization)   (None, 160, 160, 64) 256         conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_50 (Activation)      (None, 160, 160, 64) 0           bn_conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "pool1_pad (ZeroPadding2D)       (None, 162, 162, 64) 0           activation_50[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2D)  (None, 80, 80, 64)   0           pool1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2a (Conv2D)         (None, 80, 80, 64)   4160        max_pooling2d_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_51 (Activation)      (None, 80, 80, 64)   0           bn2a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_51[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_52 (Activation)      (None, 80, 80, 64)   0           bn2a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_52[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch1 (Conv2D)          (None, 80, 80, 256)  16640       max_pooling2d_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch1 (BatchNormalizatio (None, 80, 80, 256)  1024        res2a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_17 (Add)                    (None, 80, 80, 256)  0           bn2a_branch2c[0][0]              \n",
      "                                                                 bn2a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_53 (Activation)      (None, 80, 80, 256)  0           add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2a (Conv2D)         (None, 80, 80, 64)   16448       activation_53[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_54 (Activation)      (None, 80, 80, 64)   0           bn2b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_54[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_55 (Activation)      (None, 80, 80, 64)   0           bn2b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_55[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_18 (Add)                    (None, 80, 80, 256)  0           bn2b_branch2c[0][0]              \n",
      "                                                                 activation_53[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_56 (Activation)      (None, 80, 80, 256)  0           add_18[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2a (Conv2D)         (None, 80, 80, 64)   16448       activation_56[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2a (BatchNormalizati (None, 80, 80, 64)   256         res2c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_57 (Activation)      (None, 80, 80, 64)   0           bn2c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2b (Conv2D)         (None, 80, 80, 64)   36928       activation_57[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2b (BatchNormalizati (None, 80, 80, 64)   256         res2c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_58 (Activation)      (None, 80, 80, 64)   0           bn2c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2c (Conv2D)         (None, 80, 80, 256)  16640       activation_58[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2c (BatchNormalizati (None, 80, 80, 256)  1024        res2c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_19 (Add)                    (None, 80, 80, 256)  0           bn2c_branch2c[0][0]              \n",
      "                                                                 activation_56[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_59 (Activation)      (None, 80, 80, 256)  0           add_19[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2a (Conv2D)         (None, 40, 40, 128)  32896       activation_59[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_60 (Activation)      (None, 40, 40, 128)  0           bn3a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_60[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_61 (Activation)      (None, 40, 40, 128)  0           bn3a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_61[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch1 (Conv2D)          (None, 40, 40, 512)  131584      activation_59[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch1 (BatchNormalizatio (None, 40, 40, 512)  2048        res3a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_20 (Add)                    (None, 40, 40, 512)  0           bn3a_branch2c[0][0]              \n",
      "                                                                 bn3a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_62 (Activation)      (None, 40, 40, 512)  0           add_20[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_62[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_63 (Activation)      (None, 40, 40, 128)  0           bn3b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_63[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_64 (Activation)      (None, 40, 40, 128)  0           bn3b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_64[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_21 (Add)                    (None, 40, 40, 512)  0           bn3b_branch2c[0][0]              \n",
      "                                                                 activation_62[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_65 (Activation)      (None, 40, 40, 512)  0           add_21[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_65[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_66 (Activation)      (None, 40, 40, 128)  0           bn3c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_66[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_67 (Activation)      (None, 40, 40, 128)  0           bn3c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_67[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_22 (Add)                    (None, 40, 40, 512)  0           bn3c_branch2c[0][0]              \n",
      "                                                                 activation_65[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_68 (Activation)      (None, 40, 40, 512)  0           add_22[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2a (Conv2D)         (None, 40, 40, 128)  65664       activation_68[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2a (BatchNormalizati (None, 40, 40, 128)  512         res3d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_69 (Activation)      (None, 40, 40, 128)  0           bn3d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2b (Conv2D)         (None, 40, 40, 128)  147584      activation_69[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2b (BatchNormalizati (None, 40, 40, 128)  512         res3d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_70 (Activation)      (None, 40, 40, 128)  0           bn3d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2c (Conv2D)         (None, 40, 40, 512)  66048       activation_70[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2c (BatchNormalizati (None, 40, 40, 512)  2048        res3d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_23 (Add)                    (None, 40, 40, 512)  0           bn3d_branch2c[0][0]              \n",
      "                                                                 activation_68[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_71 (Activation)      (None, 40, 40, 512)  0           add_23[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2a (Conv2D)         (None, 20, 20, 256)  131328      activation_71[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_72 (Activation)      (None, 20, 20, 256)  0           bn4a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_72[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_73 (Activation)      (None, 20, 20, 256)  0           bn4a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_73[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch1 (Conv2D)          (None, 20, 20, 1024) 525312      activation_71[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch1 (BatchNormalizatio (None, 20, 20, 1024) 4096        res4a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_24 (Add)                    (None, 20, 20, 1024) 0           bn4a_branch2c[0][0]              \n",
      "                                                                 bn4a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_74 (Activation)      (None, 20, 20, 1024) 0           add_24[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_74[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_75 (Activation)      (None, 20, 20, 256)  0           bn4b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_75[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_76 (Activation)      (None, 20, 20, 256)  0           bn4b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_76[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_25 (Add)                    (None, 20, 20, 1024) 0           bn4b_branch2c[0][0]              \n",
      "                                                                 activation_74[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_77 (Activation)      (None, 20, 20, 1024) 0           add_25[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_77[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_78 (Activation)      (None, 20, 20, 256)  0           bn4c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_78[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_79 (Activation)      (None, 20, 20, 256)  0           bn4c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_79[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_26 (Add)                    (None, 20, 20, 1024) 0           bn4c_branch2c[0][0]              \n",
      "                                                                 activation_77[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_80 (Activation)      (None, 20, 20, 1024) 0           add_26[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_80[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_81 (Activation)      (None, 20, 20, 256)  0           bn4d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_81[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_82 (Activation)      (None, 20, 20, 256)  0           bn4d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_82[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_27 (Add)                    (None, 20, 20, 1024) 0           bn4d_branch2c[0][0]              \n",
      "                                                                 activation_80[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_83 (Activation)      (None, 20, 20, 1024) 0           add_27[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_83[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4e_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_84 (Activation)      (None, 20, 20, 256)  0           bn4e_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_84[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4e_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_85 (Activation)      (None, 20, 20, 256)  0           bn4e_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_85[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4e_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_28 (Add)                    (None, 20, 20, 1024) 0           bn4e_branch2c[0][0]              \n",
      "                                                                 activation_83[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_86 (Activation)      (None, 20, 20, 1024) 0           add_28[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2a (Conv2D)         (None, 20, 20, 256)  262400      activation_86[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2a (BatchNormalizati (None, 20, 20, 256)  1024        res4f_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_87 (Activation)      (None, 20, 20, 256)  0           bn4f_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2b (Conv2D)         (None, 20, 20, 256)  590080      activation_87[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2b (BatchNormalizati (None, 20, 20, 256)  1024        res4f_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_88 (Activation)      (None, 20, 20, 256)  0           bn4f_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2c (Conv2D)         (None, 20, 20, 1024) 263168      activation_88[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2c (BatchNormalizati (None, 20, 20, 1024) 4096        res4f_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_29 (Add)                    (None, 20, 20, 1024) 0           bn4f_branch2c[0][0]              \n",
      "                                                                 activation_86[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_89 (Activation)      (None, 20, 20, 1024) 0           add_29[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2a (Conv2D)         (None, 10, 10, 512)  524800      activation_89[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_90 (Activation)      (None, 10, 10, 512)  0           bn5a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_90[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_91 (Activation)      (None, 10, 10, 512)  0           bn5a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_91[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch1 (Conv2D)          (None, 10, 10, 2048) 2099200     activation_89[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch1 (BatchNormalizatio (None, 10, 10, 2048) 8192        res5a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_30 (Add)                    (None, 10, 10, 2048) 0           bn5a_branch2c[0][0]              \n",
      "                                                                 bn5a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_92 (Activation)      (None, 10, 10, 2048) 0           add_30[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2a (Conv2D)         (None, 10, 10, 512)  1049088     activation_92[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_93 (Activation)      (None, 10, 10, 512)  0           bn5b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_93[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_94 (Activation)      (None, 10, 10, 512)  0           bn5b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_94[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_31 (Add)                    (None, 10, 10, 2048) 0           bn5b_branch2c[0][0]              \n",
      "                                                                 activation_92[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_95 (Activation)      (None, 10, 10, 2048) 0           add_31[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2a (Conv2D)         (None, 10, 10, 512)  1049088     activation_95[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2a (BatchNormalizati (None, 10, 10, 512)  2048        res5c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_96 (Activation)      (None, 10, 10, 512)  0           bn5c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2b (Conv2D)         (None, 10, 10, 512)  2359808     activation_96[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2b (BatchNormalizati (None, 10, 10, 512)  2048        res5c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_97 (Activation)      (None, 10, 10, 512)  0           bn5c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2c (Conv2D)         (None, 10, 10, 2048) 1050624     activation_97[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2c (BatchNormalizati (None, 10, 10, 2048) 8192        res5c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_32 (Add)                    (None, 10, 10, 2048) 0           bn5c_branch2c[0][0]              \n",
      "                                                                 activation_95[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_98 (Activation)      (None, 10, 10, 2048) 0           add_32[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d_2 (Glo (None, 2048)         0           activation_98[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 2048)         0           global_average_pooling2d_2[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 2048)         4196352     dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_4 (Dropout)             (None, 2048)         0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "final_output (Dense)            (None, 5)            10245       dropout_4[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 27,794,309\n",
      "Trainable params: 27,741,189\n",
      "Non-trainable params: 53,120\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "for layer in model.layers:\n",
    "    layer.trainable = True\n",
    "\n",
    "es = EarlyStopping(monitor='val_loss', mode='min', patience=ES_PATIENCE, restore_best_weights=True, verbose=1)\n",
    "rlrop = ReduceLROnPlateau(monitor='val_loss', mode='min', patience=RLROP_PATIENCE, factor=DECAY_DROP, min_lr=1e-6, verbose=1)\n",
    "\n",
    "callback_list = [es, rlrop]\n",
    "optimizer = optimizers.Adam(lr=LEARNING_RATE)\n",
    "model.compile(optimizer=optimizer, loss=\"binary_crossentropy\",  metrics=['accuracy'])\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 756
    },
    "colab_type": "code",
    "id": "caeqUZYJWQ6k",
    "outputId": "8c65ebfb-4637-4fed-d4dc-c76b495910d3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "293/293 [==============================] - 268s 914ms/step - loss: 0.2535 - accuracy: 0.8959 - val_loss: 0.4255 - val_accuracy: 0.7962\n",
      "Epoch 2/40\n",
      "293/293 [==============================] - 246s 841ms/step - loss: 0.1998 - accuracy: 0.9152 - val_loss: 0.5757 - val_accuracy: 0.8055\n",
      "Epoch 3/40\n",
      "293/293 [==============================] - 246s 841ms/step - loss: 0.1823 - accuracy: 0.9218 - val_loss: 0.2257 - val_accuracy: 0.9068\n",
      "Epoch 4/40\n",
      "293/293 [==============================] - 247s 842ms/step - loss: 0.1657 - accuracy: 0.9324 - val_loss: 0.0892 - val_accuracy: 0.9161\n",
      "Epoch 5/40\n",
      "293/293 [==============================] - 247s 843ms/step - loss: 0.1496 - accuracy: 0.9362 - val_loss: 0.2838 - val_accuracy: 0.9203\n",
      "Epoch 6/40\n",
      "293/293 [==============================] - 247s 843ms/step - loss: 0.1301 - accuracy: 0.9437 - val_loss: 0.3107 - val_accuracy: 0.8711\n",
      "Epoch 7/40\n",
      "293/293 [==============================] - 247s 842ms/step - loss: 0.1214 - accuracy: 0.9516 - val_loss: 0.2837 - val_accuracy: 0.9012\n",
      "\n",
      "Epoch 00007: ReduceLROnPlateau reducing learning rate to 4.999999873689376e-05.\n",
      "Epoch 8/40\n",
      "293/293 [==============================] - 247s 844ms/step - loss: 0.0759 - accuracy: 0.9701 - val_loss: 0.0417 - val_accuracy: 0.9265\n",
      "Epoch 9/40\n",
      "293/293 [==============================] - 248s 846ms/step - loss: 0.0550 - accuracy: 0.9795 - val_loss: 0.0805 - val_accuracy: 0.9341\n",
      "Epoch 10/40\n",
      "293/293 [==============================] - 247s 842ms/step - loss: 0.0436 - accuracy: 0.9846 - val_loss: 0.5192 - val_accuracy: 0.9255\n",
      "Epoch 11/40\n",
      "293/293 [==============================] - 247s 842ms/step - loss: 0.0388 - accuracy: 0.9887 - val_loss: 0.0539 - val_accuracy: 0.9355\n",
      "\n",
      "Epoch 00011: ReduceLROnPlateau reducing learning rate to 2.499999936844688e-05.\n",
      "Epoch 12/40\n",
      "293/293 [==============================] - 248s 845ms/step - loss: 0.0259 - accuracy: 0.9920 - val_loss: 0.4038 - val_accuracy: 0.9199\n",
      "Epoch 13/40\n",
      "293/293 [==============================] - 247s 843ms/step - loss: 0.0225 - accuracy: 0.9938 - val_loss: 8.4704e-04 - val_accuracy: 0.9414\n",
      "Epoch 14/40\n",
      "293/293 [==============================] - 247s 841ms/step - loss: 0.0175 - accuracy: 0.9942 - val_loss: 0.4052 - val_accuracy: 0.9331\n",
      "Epoch 15/40\n",
      "293/293 [==============================] - 247s 844ms/step - loss: 0.0140 - accuracy: 0.9951 - val_loss: 0.1702 - val_accuracy: 0.9355\n",
      "Epoch 16/40\n",
      "293/293 [==============================] - 248s 845ms/step - loss: 0.0134 - accuracy: 0.9952 - val_loss: 0.5615 - val_accuracy: 0.9321\n",
      "\n",
      "Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.249999968422344e-05.\n",
      "Epoch 17/40\n",
      "293/293 [==============================] - 247s 844ms/step - loss: 0.0132 - accuracy: 0.9955 - val_loss: 0.0080 - val_accuracy: 0.9314\n",
      "Epoch 18/40\n",
      "293/293 [==============================] - 248s 846ms/step - loss: 0.0115 - accuracy: 0.9963 - val_loss: 0.2255 - val_accuracy: 0.9321\n",
      "Restoring model weights from the end of the best epoch\n",
      "Epoch 00018: early stopping\n"
     ]
    }
   ],
   "source": [
    "history_finetunning = model.fit_generator(generator=train_generator,\n",
    "                                          steps_per_epoch=STEP_SIZE_TRAIN,\n",
    "                                          validation_data=valid_generator,\n",
    "                                          validation_steps=STEP_SIZE_VALID,\n",
    "                                          epochs=EPOCHS,\n",
    "                                          callbacks=callback_list,\n",
    "                                          verbose=1).history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 369
    },
    "colab_type": "code",
    "id": "UcLxU0ZhWQ6m",
    "outputId": "3b67a6d9-2cc8-4753-8536-ef93939ed684"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ref - https://stackoverflow.com/questions/29188757/matplotlib-specify-format-of-floats-for-tick-lables\n",
    "plt.figure(figsize=(8,5))\n",
    "\n",
    "plt.plot(history_finetunning['accuracy'])\n",
    "plt.plot(history_finetunning['val_accuracy'])\n",
    "plt.title('Model Accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Validation'], loc='upper left')\n",
    "plt.gca().ticklabel_format(axis='both', style='plain', useOffset=False)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Cu1geHQHrINJ"
   },
   "source": [
    "### <font color='red'> 9.4 Generate Train Predictions on complete Train Data </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "i4isFqHfrIcl",
    "outputId": "d135af2f-266d-41b0-c438-c9e68290f02d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2929 validated image filenames.\n",
      "2929/2929 [==============================] - 124s 42ms/step\n"
     ]
    }
   ],
   "source": [
    "complete_datagen = ImageDataGenerator(rescale=1./255)\n",
    "complete_generator = complete_datagen.flow_from_dataframe(dataframe=df_train_train,\n",
    "                                                          directory = \"./train_images_resized_preprocessed/\",\n",
    "                                                          x_col=\"file_name\",\n",
    "                                                          target_size=(HEIGHT, WIDTH),\n",
    "                                                          batch_size=1,\n",
    "                                                          shuffle=False,\n",
    "                                                          class_mode=None)\n",
    "\n",
    "STEP_SIZE_COMPLETE = complete_generator.n//complete_generator.batch_size\n",
    "train_preds = model.predict_generator(complete_generator, steps=STEP_SIZE_COMPLETE,verbose = 1)\n",
    "train_preds = [np.argmax(pred) for pred in train_preds]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "y5coP6NwsHdS",
    "outputId": "8d7348f1-698d-4352-ba68-bba45fca9663"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Cohen Kappa score: 0.980\n",
      "Train Accuracy score : 0.965\n"
     ]
    }
   ],
   "source": [
    "print(\"Train Cohen Kappa score: %.3f\" % cohen_kappa_score(train_preds, df_train_train['diagnosis'].astype('int'), weights='quadratic'))\n",
    "print(\"Train Accuracy score : %.3f\" % accuracy_score(df_train_train['diagnosis'].astype('int'),train_preds))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ESdNiiRPWQ6o"
   },
   "source": [
    "### <font color='red'> 9.5 Evaluate Model on Test Data </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "vBjYmJDxnguQ",
    "outputId": "534699c0-d5a3-40b6-8039-1b5d0ff81b66"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "733/733 [==============================] - 456s 622ms/step\n"
     ]
    }
   ],
   "source": [
    "test_generator.reset()\n",
    "STEP_SIZE_TEST = test_generator.n//test_generator.batch_size\n",
    "test_preds = model.predict_generator(test_generator, steps=STEP_SIZE_TEST,verbose = 1)\n",
    "test_labels = [np.argmax(pred) for pred in test_preds]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qk-D7ZIvoEQ9"
   },
   "outputs": [],
   "source": [
    "def plot_conf_matrix(true,pred,classes):\n",
    "    cf = confusion_matrix(true, pred)\n",
    "    \n",
    "    df_cm = pd.DataFrame(cf, range(len(classes)), range(len(classes)))\n",
    "    plt.figure(figsize=(8,5.5))\n",
    "    sns.set(font_scale=1.4)\n",
    "    sns.heatmap(df_cm, annot=True, annot_kws={\"size\": 16},xticklabels = classes ,yticklabels = classes,fmt='g')\n",
    "    #sns.heatmap(df_cm, annot=True, annot_kws={\"size\": 16})\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 491
    },
    "colab_type": "code",
    "id": "8sxcjGJ1oR7f",
    "outputId": "029b6c98-b913-4075-f4a1-13122acaf5ae"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels = ['0 - No DR', '1 - Mild', '2 - Moderate', '3 - Severe', '4 - Proliferative DR']\n",
    "plot_conf_matrix(list(df_train_test['diagnosis'].astype(int)),test_labels,labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 436
    },
    "colab_type": "code",
    "id": "23o5zz0XWQ6r",
    "outputId": "37e5c873-fad8-4a5b-f272-1199944c9c7c"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = confusion_matrix(df_train_test['diagnosis'].astype('int'), test_labels)\n",
    "cnf_matrix_norm = cnf_matrix.astype('float') / cnf_matrix.sum(axis=1)[:, np.newaxis]\n",
    "df_cm = pd.DataFrame(cnf_matrix_norm, index=labels, columns=labels)\n",
    "plt.figure(figsize=(16, 7))\n",
    "sns.heatmap(df_cm, annot=True, fmt='.2f', cmap=\"Blues\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "ZViH2sI0WQ6v",
    "outputId": "0204531e-ec13-4c63-c7eb-1e12d6391233"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Cohen Kappa score: 0.904\n",
      "Test Accuracy score : 0.839\n"
     ]
    }
   ],
   "source": [
    "print(\"Test Cohen Kappa score: %.3f\" % cohen_kappa_score(test_labels, df_train_test['diagnosis'].astype('int'), weights='quadratic'))\n",
    "print(\"Test Accuracy score : %.3f\" % accuracy_score(df_train_test['diagnosis'].astype('int'),test_labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wrNLulwLWQ62"
   },
   "source": [
    "### <font color='red'> 9.6 ResNet50 Models Summary </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 118
    },
    "colab_type": "code",
    "id": "ncmJ7qMzWQ62",
    "outputId": "2f860351-5842-4a19-d260-d27da55b7954"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+----------------------------+----------------------------+----------------------+-------------------------------+-----------+----------+\n",
      "| S.No. |       ResNet50 Model       |      Image Processing      |  Data Augmentation   | Hyperparameters(BS,Opt,lr,ep) | Train QWK | Test QWK |\n",
      "+-------+----------------------------+----------------------------+----------------------+-------------------------------+-----------+----------+\n",
      "|   1   | R-P-D-p(0.5)-D-p(0.5)-S(5) |             --             | Hor Flip,Scale 1/255 |      (4,'Adam','1e-4',7)      |   0.912   |  0.905   |\n",
      "|   2   | R-P-D-p(0.5)-D-p(0.5)-S(5) | Circle Crop, Gaussian Blur | Hor Flip,Scale 1/255 |      (4,'Adam','1e-4',7)      |    0.98   |  0.904   |\n",
      "+-------+----------------------------+----------------------------+----------------------+-------------------------------+-----------+----------+\n"
     ]
    }
   ],
   "source": [
    "x = PrettyTable()\n",
    "x.field_names = [\"S.No.\",\"ResNet50 Model\",\"Image Processing\",\"Data Augmentation\",\"Hyperparameters(BS,Opt,lr,ep)\",\"Train QWK\",\"Test QWK\"]\n",
    "\n",
    "x.add_row([1,\"R-P-D-p(0.5)-D-p(0.5)-S(5)\",\"--\",\"Hor Flip,Scale 1/255\",\"(4,'Adam','1e-4',7)\",\"0.912\",\"0.905\"])\n",
    "x.add_row([2,\"R-P-D-p(0.5)-D-p(0.5)-S(5)\",\"Circle Crop, Gaussian Blur\",\"Hor Flip,Scale 1/255\",\"(4,'Adam','1e-4',7)\",\"0.98\",\"0.904\"])\n",
    "\n",
    "print(x)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "3_resnet50(colab).ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}