nc-severstal04-classification-vgg-4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:17.812717Z",
     "iopub.status.busy": "2021-03-19T12:08:17.812005Z",
     "iopub.status.idle": "2021-03-19T12:08:24.077151Z",
     "shell.execute_reply": "2021-03-19T12:08:24.077737Z"
    },
    "papermill": {
     "duration": 6.291746,
     "end_time": "2021-03-19T12:08:24.077956",
     "exception": false,
     "start_time": "2021-03-19T12:08:17.786210",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import cv2\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.callbacks import Callback, ModelCheckpoint, ReduceLROnPlateau,EarlyStopping\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.models import Sequential, Model, load_model\n",
    "from tensorflow.keras.layers import Input, Dense, Conv2D, MaxPooling2D\n",
    "from tensorflow.keras.layers import GlobalAveragePooling2D, AveragePooling2D\n",
    "from tensorflow.keras.layers import ZeroPadding2D, Dropout, Flatten, add, Reshape\n",
    "from tensorflow.keras.layers import Activation, BatchNormalization\n",
    "from tensorflow.keras.optimizers import Adam, SGD, Nadam\n",
    "from tensorflow.keras.utils import plot_model,to_categorical\n",
    "from tensorflow.keras import backend as K\n",
    "from tensorflow.keras import initializers\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.applications import DenseNet121\n",
    "from tqdm import tqdm\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from tensorflow.keras.applications.vgg16 import preprocess_input\n",
    "from tensorflow.keras.applications.vgg16 import decode_predictions\n",
    "from tensorflow.keras.applications.vgg16 import VGG16"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.019574,
     "end_time": "2021-03-19T12:08:24.117547",
     "exception": false,
     "start_time": "2021-03-19T12:08:24.097973",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Read train and test files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:24.188865Z",
     "iopub.status.busy": "2021-03-19T12:08:24.187635Z",
     "iopub.status.idle": "2021-03-19T12:08:24.773653Z",
     "shell.execute_reply": "2021-03-19T12:08:24.774397Z"
    },
    "papermill": {
     "duration": 0.635475,
     "end_time": "2021-03-19T12:08:24.774602",
     "exception": false,
     "start_time": "2021-03-19T12:08:24.139127",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7095, 3)\n"
     ]
    }
   ],
   "source": [
    "train_df = pd.read_csv('../input/severstal-steel-defect-detection/train.csv')\n",
    "\n",
    "print(train_df.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:24.818559Z",
     "iopub.status.busy": "2021-03-19T12:08:24.817515Z",
     "iopub.status.idle": "2021-03-19T12:08:24.825726Z",
     "shell.execute_reply": "2021-03-19T12:08:24.826147Z"
    },
    "papermill": {
     "duration": 0.036796,
     "end_time": "2021-03-19T12:08:24.826252",
     "exception": false,
     "start_time": "2021-03-19T12:08:24.789456",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "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>ImageId</th>\n",
       "      <th>ClassId</th>\n",
       "      <th>EncodedPixels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0002cc93b.jpg</td>\n",
       "      <td>1</td>\n",
       "      <td>29102 12 29346 24 29602 24 29858 24 30114 24 3...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0007a71bf.jpg</td>\n",
       "      <td>3</td>\n",
       "      <td>18661 28 18863 82 19091 110 19347 110 19603 11...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>000a4bcdd.jpg</td>\n",
       "      <td>1</td>\n",
       "      <td>37607 3 37858 8 38108 14 38359 20 38610 25 388...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>000f6bf48.jpg</td>\n",
       "      <td>4</td>\n",
       "      <td>131973 1 132228 4 132483 6 132738 8 132993 11 ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0014fce06.jpg</td>\n",
       "      <td>3</td>\n",
       "      <td>229501 11 229741 33 229981 55 230221 77 230468...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ImageId  ClassId                                      EncodedPixels\n",
       "0  0002cc93b.jpg        1  29102 12 29346 24 29602 24 29858 24 30114 24 3...\n",
       "1  0007a71bf.jpg        3  18661 28 18863 82 19091 110 19347 110 19603 11...\n",
       "2  000a4bcdd.jpg        1  37607 3 37858 8 38108 14 38359 20 38610 25 388...\n",
       "3  000f6bf48.jpg        4  131973 1 132228 4 132483 6 132738 8 132993 11 ...\n",
       "4  0014fce06.jpg        3  229501 11 229741 33 229981 55 230221 77 230468..."
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.013532,
     "end_time": "2021-03-19T12:08:24.854540",
     "exception": false,
     "start_time": "2021-03-19T12:08:24.841008",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Combine defected and no-defect images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:24.891156Z",
     "iopub.status.busy": "2021-03-19T12:08:24.890427Z",
     "iopub.status.idle": "2021-03-19T12:08:26.939152Z",
     "shell.execute_reply": "2021-03-19T12:08:26.939985Z"
    },
    "papermill": {
     "duration": 2.071679,
     "end_time": "2021-03-19T12:08:26.940152",
     "exception": false,
     "start_time": "2021-03-19T12:08:24.868473",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(12568, 2)\n",
      "1    6666\n",
      "0    5902\n",
      "Name: defect_label, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Data Pre-Processing\n",
    "TRAIN_PATH = '../input/severstal-steel-defect-detection/train_images/'\n",
    "TEST_PATH = '../input/severstal-steel-defect-detection/test_images/'\n",
    "\n",
    "train_flist = os.listdir(TRAIN_PATH) \n",
    "\n",
    "image_defects = train_df['ImageId'].unique().tolist()\n",
    "image_no_defects=[item for item in train_flist if item not in image_defects]\n",
    "\n",
    "#df_bin is the dataframe with label defect based on train_df \n",
    "df_bin = train_df.drop_duplicates(subset=['ImageId'], keep='last').copy()\n",
    "df_bin.loc[:, 'defect_label'] = 1\n",
    "\n",
    "df_noDefect_bin = pd.DataFrame(image_no_defects,columns=['ImageId'])\n",
    "df_noDefect_bin.loc[:, 'defect_label'] = 0\n",
    "\n",
    "frames = [df_bin[['ImageId', \"defect_label\"]], df_noDefect_bin]\n",
    "df = pd.concat(frames).reset_index(drop=True)\n",
    "print(df.shape)\n",
    "print(df['defect_label'].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:26.979871Z",
     "iopub.status.busy": "2021-03-19T12:08:26.979155Z",
     "iopub.status.idle": "2021-03-19T12:08:26.982823Z",
     "shell.execute_reply": "2021-03-19T12:08:26.983234Z"
    },
    "papermill": {
     "duration": 0.027148,
     "end_time": "2021-03-19T12:08:26.983338",
     "exception": false,
     "start_time": "2021-03-19T12:08:26.956190",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "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>ImageId</th>\n",
       "      <th>defect_label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0002cc93b.jpg</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ImageId  defect_label\n",
       "0  0002cc93b.jpg             1"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.014427,
     "end_time": "2021-03-19T12:08:27.012285",
     "exception": false,
     "start_time": "2021-03-19T12:08:26.997858",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Split train data, load and resave images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:08:27.057384Z",
     "iopub.status.busy": "2021-03-19T12:08:27.056767Z",
     "iopub.status.idle": "2021-03-19T12:12:20.952645Z",
     "shell.execute_reply": "2021-03-19T12:12:20.952151Z"
    },
    "papermill": {
     "duration": 233.925629,
     "end_time": "2021-03-19T12:12:20.952758",
     "exception": false,
     "start_time": "2021-03-19T12:08:27.027129",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 6597/6597 [02:01<00:00, 54.51it/s]\n",
      "100%|██████████| 2200/2200 [00:40<00:00, 54.77it/s]\n",
      "100%|██████████| 3771/3771 [01:12<00:00, 51.91it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6597, 2) (2200, 2) (3771, 2)\n",
      "1    3499\n",
      "0    3098\n",
      "Name: defect_label, dtype: int64\n",
      "1    1167\n",
      "0    1033\n",
      "Name: defect_label, dtype: int64\n",
      "1    2000\n",
      "0    1771\n",
      "Name: defect_label, dtype: int64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:43: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n"
     ]
    }
   ],
   "source": [
    "nb_images = df.shape[0]\n",
    "df_train, df_test = train_test_split(df, test_size=0.3, \n",
    "                                                    stratify = df['defect_label'], \n",
    "                                                    random_state=2021)\n",
    "\n",
    "df_train, df_val = train_test_split(df_train, test_size=0.25, \n",
    "                                                  stratify = df_train['defect_label'],\n",
    "                                                  random_state=2021) \n",
    "\n",
    "# 0.25 x 0.8 = 0.2\n",
    "\n",
    "IMAGE_SIZE = 224\n",
    "\n",
    "def load_img(code, base, resize=True):\n",
    "    path = f'{base}/{code}'\n",
    "    img = cv2.imread(path)\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    if resize:\n",
    "        img = cv2.resize(img, (IMAGE_SIZE, IMAGE_SIZE))\n",
    "    return img\n",
    "\n",
    "def validate_path(path):\n",
    "    if not os.path.exists(path):\n",
    "        os.makedirs(path)\n",
    "\n",
    "train_path = '../tmp/train'\n",
    "validate_path(train_path)\n",
    "val_path = '../tmp/val'\n",
    "validate_path(val_path)\n",
    "test_path = '../tmp/test'\n",
    "validate_path(test_path)\n",
    "\n",
    "def process_img(df, train_path):\n",
    "    #read image, resize and save\n",
    "    for code in tqdm(df['ImageId']):\n",
    "        img = load_img(\n",
    "            code,\n",
    "            base='../input/severstal-steel-defect-detection/train_images'\n",
    "        )\n",
    "        path = code.replace('.jpg', '')\n",
    "        cv2.imwrite(f'{train_path}/{path}.png', img)\n",
    "\n",
    "    df['ImageId'] = df['ImageId'].apply(lambda x: x.replace('.jpg', '.png'))\n",
    "    return df\n",
    "df_train = process_img(df_train,train_path)\n",
    "\n",
    "df_val = process_img(df_val,val_path)\n",
    "\n",
    "df_test = process_img(df_test,test_path)\n",
    "\n",
    "print(df_train.shape,df_val.shape,df_test.shape)\n",
    "print(df_train['defect_label'].value_counts())\n",
    "print(df_val['defect_label'].value_counts())\n",
    "print(df_test['defect_label'].value_counts())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.590493,
     "end_time": "2021-03-19T12:12:22.119755",
     "exception": false,
     "start_time": "2021-03-19T12:12:21.529262",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Define ImageDataGenerator and data augmentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:12:23.317129Z",
     "iopub.status.busy": "2021-03-19T12:12:23.315746Z",
     "iopub.status.idle": "2021-03-19T12:12:23.423395Z",
     "shell.execute_reply": "2021-03-19T12:12:23.424155Z"
    },
    "papermill": {
     "duration": 0.720726,
     "end_time": "2021-03-19T12:12:23.424349",
     "exception": false,
     "start_time": "2021-03-19T12:12:22.703623",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 6597 validated image filenames.\n",
      "Found 2200 validated image filenames.\n",
      "Found 3771 validated image filenames.\n"
     ]
    }
   ],
   "source": [
    "BATCH_SIZE = 32\n",
    "\n",
    "# Using original generator\n",
    "data_generator = ImageDataGenerator(\n",
    "        zoom_range=0.1,  # set range for random zoom\n",
    "        # set mode for filling points outside the input boundaries\n",
    "        fill_mode='constant',\n",
    "        cval=0.,\n",
    "        rotation_range=10,\n",
    "        height_shift_range=0.05,\n",
    "        width_shift_range=0.05,\n",
    "        horizontal_flip=True,\n",
    "        vertical_flip=True,\n",
    "        rescale=1/255.,\n",
    "    )\n",
    "val_generator = ImageDataGenerator(\n",
    "        rescale=1/255.\n",
    "    )\n",
    "\n",
    "train_gen = data_generator.flow_from_dataframe(\n",
    "        df_train, \n",
    "        directory='../tmp/train',\n",
    "        x_col='ImageId', \n",
    "        y_col='defect_label', \n",
    "        class_mode='raw',\n",
    "        target_size=(IMAGE_SIZE, IMAGE_SIZE),\n",
    "        batch_size=BATCH_SIZE,\n",
    "        )\n",
    "\n",
    "val_gen =  val_generator.flow_from_dataframe(\n",
    "        df_val, \n",
    "        directory='../tmp/val',\n",
    "        x_col='ImageId', \n",
    "        y_col='defect_label', \n",
    "        class_mode='raw',\n",
    "        target_size=(IMAGE_SIZE, IMAGE_SIZE),\n",
    "        batch_size=BATCH_SIZE,\n",
    "        )\n",
    "test_gen =  val_generator.flow_from_dataframe(\n",
    "        df_test, \n",
    "        directory='../tmp/test',\n",
    "        x_col='ImageId', \n",
    "        y_col='defect_label', \n",
    "        class_mode='raw',\n",
    "        target_size=(IMAGE_SIZE, IMAGE_SIZE),\n",
    "        batch_size=BATCH_SIZE,\n",
    "        shuffle=False\n",
    "        )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.59389,
     "end_time": "2021-03-19T12:12:24.602958",
     "exception": false,
     "start_time": "2021-03-19T12:12:24.009068",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Define Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:12:25.989239Z",
     "iopub.status.busy": "2021-03-19T12:12:25.987873Z",
     "iopub.status.idle": "2021-03-19T12:12:29.661378Z",
     "shell.execute_reply": "2021-03-19T12:12:29.660682Z"
    },
    "papermill": {
     "duration": 4.460607,
     "end_time": "2021-03-19T12:12:29.661545",
     "exception": false,
     "start_time": "2021-03-19T12:12:25.200938",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
      "58892288/58889256 [==============================] - 0s 0us/step\n",
      "Model: \"functional_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 224, 224, 3)]     0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 25088)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 256)               6422784   \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 257       \n",
      "=================================================================\n",
      "Total params: 21,137,729\n",
      "Trainable params: 6,423,041\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "input_shape = (IMAGE_SIZE,IMAGE_SIZE,3)\n",
    "\n",
    "def build_model():\n",
    "    \n",
    "    model = VGG16(weights='imagenet',include_top=False, input_shape=input_shape)\n",
    "    # add new classifier layers\n",
    "    for layer in model.layers:\n",
    "        layer.trainable = False\n",
    "\n",
    "    flat1 = Flatten()(model.layers[-1].output)\n",
    "    den1 = Dense(256, activation= 'relu')(flat1)\n",
    "    output = Dense(1, activation='sigmoid')(den1)\n",
    "    # define new model\n",
    "    model = Model(inputs=model.inputs, outputs=output)\n",
    "\n",
    "    model.compile(\n",
    "        loss='binary_crossentropy',\n",
    "        optimizer= Adam(learning_rate = 0.001),\n",
    "        metrics=['accuracy']\n",
    "    )\n",
    "    \n",
    "\n",
    "    return model\n",
    "model = build_model()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.599944,
     "end_time": "2021-03-19T12:12:30.864978",
     "exception": false,
     "start_time": "2021-03-19T12:12:30.265034",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Model training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:12:32.039346Z",
     "iopub.status.busy": "2021-03-19T12:12:32.038596Z",
     "iopub.status.idle": "2021-03-19T12:12:32.041419Z",
     "shell.execute_reply": "2021-03-19T12:12:32.040963Z"
    },
    "papermill": {
     "duration": 0.59569,
     "end_time": "2021-03-19T12:12:32.041509",
     "exception": false,
     "start_time": "2021-03-19T12:12:31.445819",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "epochs = 50\n",
    "\n",
    "import math\n",
    "# We take the ceiling because we do not drop the remainder of the batch\n",
    "compute_steps_per_epoch = lambda x: int(math.ceil(1. * x / BATCH_SIZE))\n",
    "\n",
    "steps_per_epoch = compute_steps_per_epoch(df_train.shape[0])\n",
    "validation_steps = compute_steps_per_epoch(df_val.shape[0])\n",
    "\n",
    "es = EarlyStopping(monitor='val_loss', mode='min', verbose=2, patience=20)\n",
    "reduce_lr = ReduceLROnPlateau(monitor='val_loss', patience=5, verbose=1, min_lr=1e-6)\n",
    "\n",
    "mc = ModelCheckpoint(\n",
    "    'model.h5', \n",
    "    monitor='val_accuracy', \n",
    "    verbose=1, \n",
    "    save_best_only=True, \n",
    "    save_weights_only=False,\n",
    "    mode='auto'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T12:12:33.202992Z",
     "iopub.status.busy": "2021-03-19T12:12:33.202248Z",
     "iopub.status.idle": "2021-03-19T13:31:26.323993Z",
     "shell.execute_reply": "2021-03-19T13:31:26.323069Z"
    },
    "papermill": {
     "duration": 4733.69914,
     "end_time": "2021-03-19T13:31:26.324111",
     "exception": false,
     "start_time": "2021-03-19T12:12:32.624971",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.6612 - accuracy: 0.6799\n",
      "Epoch 00001: val_accuracy improved from -inf to 0.69182, saving model to model.h5\n",
      "207/207 [==============================] - 94s 452ms/step - loss: 0.6612 - accuracy: 0.6799 - val_loss: 0.5413 - val_accuracy: 0.6918\n",
      "Epoch 2/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.4772 - accuracy: 0.7606\n",
      "Epoch 00002: val_accuracy improved from 0.69182 to 0.79500, saving model to model.h5\n",
      "207/207 [==============================] - 91s 440ms/step - loss: 0.4772 - accuracy: 0.7606 - val_loss: 0.4326 - val_accuracy: 0.7950\n",
      "Epoch 3/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.4579 - accuracy: 0.7693\n",
      "Epoch 00003: val_accuracy improved from 0.79500 to 0.80045, saving model to model.h5\n",
      "207/207 [==============================] - 91s 442ms/step - loss: 0.4579 - accuracy: 0.7693 - val_loss: 0.4258 - val_accuracy: 0.8005\n",
      "Epoch 4/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.4231 - accuracy: 0.7941\n",
      "Epoch 00004: val_accuracy did not improve from 0.80045\n",
      "207/207 [==============================] - 91s 439ms/step - loss: 0.4231 - accuracy: 0.7941 - val_loss: 0.4143 - val_accuracy: 0.7941\n",
      "Epoch 5/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.4260 - accuracy: 0.7910\n",
      "Epoch 00005: val_accuracy did not improve from 0.80045\n",
      "207/207 [==============================] - 90s 435ms/step - loss: 0.4260 - accuracy: 0.7910 - val_loss: 0.4730 - val_accuracy: 0.7723\n",
      "Epoch 6/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3930 - accuracy: 0.8067\n",
      "Epoch 00006: val_accuracy did not improve from 0.80045\n",
      "207/207 [==============================] - 92s 442ms/step - loss: 0.3930 - accuracy: 0.8067 - val_loss: 0.4095 - val_accuracy: 0.7982\n",
      "Epoch 7/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3734 - accuracy: 0.8255\n",
      "Epoch 00007: val_accuracy did not improve from 0.80045\n",
      "207/207 [==============================] - 91s 439ms/step - loss: 0.3734 - accuracy: 0.8255 - val_loss: 0.4295 - val_accuracy: 0.7845\n",
      "Epoch 8/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3687 - accuracy: 0.8264\n",
      "Epoch 00008: val_accuracy improved from 0.80045 to 0.82409, saving model to model.h5\n",
      "207/207 [==============================] - 91s 438ms/step - loss: 0.3687 - accuracy: 0.8264 - val_loss: 0.3748 - val_accuracy: 0.8241\n",
      "Epoch 9/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3806 - accuracy: 0.8166\n",
      "Epoch 00009: val_accuracy did not improve from 0.82409\n",
      "207/207 [==============================] - 91s 438ms/step - loss: 0.3806 - accuracy: 0.8166 - val_loss: 0.3989 - val_accuracy: 0.8164\n",
      "Epoch 10/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3607 - accuracy: 0.8325\n",
      "Epoch 00010: val_accuracy improved from 0.82409 to 0.83500, saving model to model.h5\n",
      "207/207 [==============================] - 92s 443ms/step - loss: 0.3607 - accuracy: 0.8325 - val_loss: 0.3641 - val_accuracy: 0.8350\n",
      "Epoch 11/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3539 - accuracy: 0.8354\n",
      "Epoch 00011: val_accuracy improved from 0.83500 to 0.84045, saving model to model.h5\n",
      "207/207 [==============================] - 94s 454ms/step - loss: 0.3539 - accuracy: 0.8354 - val_loss: 0.3574 - val_accuracy: 0.8405\n",
      "Epoch 12/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3528 - accuracy: 0.8404\n",
      "Epoch 00012: val_accuracy did not improve from 0.84045\n",
      "207/207 [==============================] - 93s 451ms/step - loss: 0.3528 - accuracy: 0.8404 - val_loss: 0.4055 - val_accuracy: 0.8050\n",
      "Epoch 13/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3415 - accuracy: 0.8440\n",
      "Epoch 00013: val_accuracy did not improve from 0.84045\n",
      "207/207 [==============================] - 91s 441ms/step - loss: 0.3415 - accuracy: 0.8440 - val_loss: 0.3852 - val_accuracy: 0.8064\n",
      "Epoch 14/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3334 - accuracy: 0.8484\n",
      "Epoch 00014: val_accuracy improved from 0.84045 to 0.84136, saving model to model.h5\n",
      "207/207 [==============================] - 93s 451ms/step - loss: 0.3334 - accuracy: 0.8484 - val_loss: 0.3499 - val_accuracy: 0.8414\n",
      "Epoch 15/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3351 - accuracy: 0.8443\n",
      "Epoch 00015: val_accuracy improved from 0.84136 to 0.84227, saving model to model.h5\n",
      "207/207 [==============================] - 92s 446ms/step - loss: 0.3351 - accuracy: 0.8443 - val_loss: 0.3438 - val_accuracy: 0.8423\n",
      "Epoch 16/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3256 - accuracy: 0.8510\n",
      "Epoch 00016: val_accuracy improved from 0.84227 to 0.84409, saving model to model.h5\n",
      "207/207 [==============================] - 93s 450ms/step - loss: 0.3256 - accuracy: 0.8510 - val_loss: 0.3448 - val_accuracy: 0.8441\n",
      "Epoch 17/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3207 - accuracy: 0.8539\n",
      "Epoch 00017: val_accuracy did not improve from 0.84409\n",
      "207/207 [==============================] - 92s 445ms/step - loss: 0.3207 - accuracy: 0.8539 - val_loss: 0.3527 - val_accuracy: 0.8382\n",
      "Epoch 18/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3270 - accuracy: 0.8507\n",
      "Epoch 00018: val_accuracy did not improve from 0.84409\n",
      "207/207 [==============================] - 92s 446ms/step - loss: 0.3270 - accuracy: 0.8507 - val_loss: 0.3451 - val_accuracy: 0.8441\n",
      "Epoch 19/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3123 - accuracy: 0.8605\n",
      "Epoch 00019: val_accuracy improved from 0.84409 to 0.84636, saving model to model.h5\n",
      "207/207 [==============================] - 95s 461ms/step - loss: 0.3123 - accuracy: 0.8605 - val_loss: 0.3428 - val_accuracy: 0.8464\n",
      "Epoch 20/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3219 - accuracy: 0.8521\n",
      "Epoch 00020: val_accuracy improved from 0.84636 to 0.85136, saving model to model.h5\n",
      "207/207 [==============================] - 96s 464ms/step - loss: 0.3219 - accuracy: 0.8521 - val_loss: 0.3329 - val_accuracy: 0.8514\n",
      "Epoch 21/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3123 - accuracy: 0.8599\n",
      "Epoch 00021: val_accuracy did not improve from 0.85136\n",
      "207/207 [==============================] - 96s 465ms/step - loss: 0.3123 - accuracy: 0.8599 - val_loss: 0.3559 - val_accuracy: 0.8386\n",
      "Epoch 22/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3147 - accuracy: 0.8552\n",
      "Epoch 00022: val_accuracy improved from 0.85136 to 0.85318, saving model to model.h5\n",
      "207/207 [==============================] - 97s 469ms/step - loss: 0.3147 - accuracy: 0.8552 - val_loss: 0.3276 - val_accuracy: 0.8532\n",
      "Epoch 23/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3048 - accuracy: 0.8636\n",
      "Epoch 00023: val_accuracy did not improve from 0.85318\n",
      "207/207 [==============================] - 97s 469ms/step - loss: 0.3048 - accuracy: 0.8636 - val_loss: 0.3341 - val_accuracy: 0.8514\n",
      "Epoch 24/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.3031 - accuracy: 0.8643\n",
      "Epoch 00024: val_accuracy did not improve from 0.85318\n",
      "207/207 [==============================] - 98s 473ms/step - loss: 0.3031 - accuracy: 0.8643 - val_loss: 0.3882 - val_accuracy: 0.8282\n",
      "Epoch 25/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2947 - accuracy: 0.8693\n",
      "Epoch 00025: val_accuracy improved from 0.85318 to 0.85773, saving model to model.h5\n",
      "207/207 [==============================] - 95s 460ms/step - loss: 0.2947 - accuracy: 0.8693 - val_loss: 0.3205 - val_accuracy: 0.8577\n",
      "Epoch 26/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2908 - accuracy: 0.8716\n",
      "Epoch 00026: val_accuracy did not improve from 0.85773\n",
      "207/207 [==============================] - 97s 470ms/step - loss: 0.2908 - accuracy: 0.8716 - val_loss: 0.3461 - val_accuracy: 0.8495\n",
      "Epoch 27/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2875 - accuracy: 0.8716\n",
      "Epoch 00027: val_accuracy did not improve from 0.85773\n",
      "207/207 [==============================] - 97s 469ms/step - loss: 0.2875 - accuracy: 0.8716 - val_loss: 0.3474 - val_accuracy: 0.8509\n",
      "Epoch 28/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2917 - accuracy: 0.8718\n",
      "Epoch 00028: val_accuracy improved from 0.85773 to 0.86091, saving model to model.h5\n",
      "207/207 [==============================] - 98s 474ms/step - loss: 0.2917 - accuracy: 0.8718 - val_loss: 0.3186 - val_accuracy: 0.8609\n",
      "Epoch 29/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2934 - accuracy: 0.8675\n",
      "Epoch 00029: val_accuracy did not improve from 0.86091\n",
      "207/207 [==============================] - 99s 481ms/step - loss: 0.2934 - accuracy: 0.8675 - val_loss: 0.3513 - val_accuracy: 0.8427\n",
      "Epoch 30/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2867 - accuracy: 0.8733\n",
      "Epoch 00030: val_accuracy did not improve from 0.86091\n",
      "207/207 [==============================] - 97s 469ms/step - loss: 0.2867 - accuracy: 0.8733 - val_loss: 0.3215 - val_accuracy: 0.8582\n",
      "Epoch 31/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2812 - accuracy: 0.8790\n",
      "Epoch 00031: val_accuracy did not improve from 0.86091\n",
      "207/207 [==============================] - 95s 457ms/step - loss: 0.2812 - accuracy: 0.8790 - val_loss: 0.3417 - val_accuracy: 0.8436\n",
      "Epoch 32/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2811 - accuracy: 0.8724\n",
      "Epoch 00032: val_accuracy did not improve from 0.86091\n",
      "207/207 [==============================] - 94s 453ms/step - loss: 0.2811 - accuracy: 0.8724 - val_loss: 0.3262 - val_accuracy: 0.8545\n",
      "Epoch 33/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2829 - accuracy: 0.8733\n",
      "Epoch 00033: val_accuracy improved from 0.86091 to 0.86273, saving model to model.h5\n",
      "207/207 [==============================] - 94s 455ms/step - loss: 0.2829 - accuracy: 0.8733 - val_loss: 0.3136 - val_accuracy: 0.8627\n",
      "Epoch 34/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2681 - accuracy: 0.8833\n",
      "Epoch 00034: val_accuracy did not improve from 0.86273\n",
      "207/207 [==============================] - 92s 446ms/step - loss: 0.2681 - accuracy: 0.8833 - val_loss: 0.3357 - val_accuracy: 0.8514\n",
      "Epoch 35/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2803 - accuracy: 0.8798\n",
      "Epoch 00035: val_accuracy did not improve from 0.86273\n",
      "207/207 [==============================] - 95s 460ms/step - loss: 0.2803 - accuracy: 0.8798 - val_loss: 0.3203 - val_accuracy: 0.8586\n",
      "Epoch 36/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2779 - accuracy: 0.8781\n",
      "Epoch 00036: val_accuracy improved from 0.86273 to 0.87000, saving model to model.h5\n",
      "207/207 [==============================] - 94s 455ms/step - loss: 0.2779 - accuracy: 0.8781 - val_loss: 0.3137 - val_accuracy: 0.8700\n",
      "Epoch 37/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2720 - accuracy: 0.8812\n",
      "Epoch 00037: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 91s 441ms/step - loss: 0.2720 - accuracy: 0.8812 - val_loss: 0.3478 - val_accuracy: 0.8477\n",
      "Epoch 38/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2698 - accuracy: 0.8806\n",
      "Epoch 00038: val_accuracy did not improve from 0.87000\n",
      "\n",
      "Epoch 00038: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "207/207 [==============================] - 94s 453ms/step - loss: 0.2698 - accuracy: 0.8806 - val_loss: 0.3201 - val_accuracy: 0.8532\n",
      "Epoch 39/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2441 - accuracy: 0.8978\n",
      "Epoch 00039: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 95s 460ms/step - loss: 0.2441 - accuracy: 0.8978 - val_loss: 0.3191 - val_accuracy: 0.8577\n",
      "Epoch 40/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2362 - accuracy: 0.9025\n",
      "Epoch 00040: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 92s 445ms/step - loss: 0.2362 - accuracy: 0.9025 - val_loss: 0.3215 - val_accuracy: 0.8564\n",
      "Epoch 41/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2317 - accuracy: 0.8977\n",
      "Epoch 00041: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 94s 452ms/step - loss: 0.2317 - accuracy: 0.8977 - val_loss: 0.3215 - val_accuracy: 0.8586\n",
      "Epoch 42/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2284 - accuracy: 0.9036\n",
      "Epoch 00042: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 94s 453ms/step - loss: 0.2284 - accuracy: 0.9036 - val_loss: 0.3195 - val_accuracy: 0.8600\n",
      "Epoch 43/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2324 - accuracy: 0.9000\n",
      "Epoch 00043: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 93s 447ms/step - loss: 0.2324 - accuracy: 0.9000 - val_loss: 0.3064 - val_accuracy: 0.8700\n",
      "Epoch 44/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2316 - accuracy: 0.9015\n",
      "Epoch 00044: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 95s 459ms/step - loss: 0.2316 - accuracy: 0.9015 - val_loss: 0.3143 - val_accuracy: 0.8641\n",
      "Epoch 45/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2351 - accuracy: 0.9028\n",
      "Epoch 00045: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 95s 461ms/step - loss: 0.2351 - accuracy: 0.9028 - val_loss: 0.3127 - val_accuracy: 0.8664\n",
      "Epoch 46/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2292 - accuracy: 0.9050\n",
      "Epoch 00046: val_accuracy did not improve from 0.87000\n",
      "207/207 [==============================] - 94s 456ms/step - loss: 0.2292 - accuracy: 0.9050 - val_loss: 0.3135 - val_accuracy: 0.8659\n",
      "Epoch 47/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2217 - accuracy: 0.9068\n",
      "Epoch 00047: val_accuracy improved from 0.87000 to 0.87545, saving model to model.h5\n",
      "207/207 [==============================] - 97s 467ms/step - loss: 0.2217 - accuracy: 0.9068 - val_loss: 0.3061 - val_accuracy: 0.8755\n",
      "Epoch 48/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2290 - accuracy: 0.9045\n",
      "Epoch 00048: val_accuracy did not improve from 0.87545\n",
      "207/207 [==============================] - 97s 467ms/step - loss: 0.2290 - accuracy: 0.9045 - val_loss: 0.3107 - val_accuracy: 0.8727\n",
      "Epoch 49/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2272 - accuracy: 0.9077\n",
      "Epoch 00049: val_accuracy did not improve from 0.87545\n",
      "207/207 [==============================] - 94s 454ms/step - loss: 0.2272 - accuracy: 0.9077 - val_loss: 0.3125 - val_accuracy: 0.8709\n",
      "Epoch 50/50\n",
      "207/207 [==============================] - ETA: 0s - loss: 0.2262 - accuracy: 0.9034\n",
      "Epoch 00050: val_accuracy did not improve from 0.87545\n",
      "207/207 [==============================] - 96s 464ms/step - loss: 0.2262 - accuracy: 0.9034 - val_loss: 0.3100 - val_accuracy: 0.8718\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    train_gen,\n",
    "    steps_per_epoch= steps_per_epoch,\n",
    "    validation_data=val_gen,\n",
    "    validation_steps= validation_steps,\n",
    "    epochs=epochs,\n",
    "    callbacks=[es,mc,reduce_lr]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T13:31:33.635321Z",
     "iopub.status.busy": "2021-03-19T13:31:33.632250Z",
     "iopub.status.idle": "2021-03-19T13:31:33.938373Z",
     "shell.execute_reply": "2021-03-19T13:31:33.938850Z"
    },
    "papermill": {
     "duration": 3.789021,
     "end_time": "2021-03-19T13:31:33.938979",
     "exception": false,
     "start_time": "2021-03-19T13:31:30.149958",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x1152 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def history_plot(model_id, history1):\n",
    "\tplt.figure(figsize=(14,16)) \n",
    "\t# plot loss\n",
    "\tplt.subplot(2, 1, 1)\n",
    "\tplt.title('Cross Entropy Loss - ' + model_id, fontsize = 12)\n",
    "\tplt.plot(history1.history['loss'], color='blue', label='train base')\n",
    "\tplt.plot(history1.history['val_loss'], color='cyan', label='val base')\n",
    "\tplt.xlabel(\"Epochs\",fontsize = 12)\n",
    "\tplt.ylabel(\"Loss\", fontsize = 12)\n",
    "\tplt.legend(loc='upper right')\n",
    "\n",
    "\t# plot accuracy\n",
    "\tplt.subplot(2, 1, 2)\n",
    "\tplt.title('Classification Accuracy ' + model_id, fontsize = 10)\n",
    "\tplt.plot(history1.history['accuracy'], color='blue', label='train base')\n",
    "\tplt.plot(history1.history['val_accuracy'], color='cyan', label='val base')\n",
    "\tplt.xlabel(\"Epochs\", fontsize = 12)\n",
    "\tplt.ylabel(\"Accuracy\", fontsize = 12)\n",
    "\tplt.legend(loc='lower right')\n",
    "\n",
    "\tplt.show()\n",
    "\n",
    "print(history.history.keys())\n",
    "\n",
    "history_plot('CNN Model', history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T13:31:41.265069Z",
     "iopub.status.busy": "2021-03-19T13:31:41.237805Z",
     "iopub.status.idle": "2021-03-19T13:31:56.069792Z",
     "shell.execute_reply": "2021-03-19T13:31:56.068719Z"
    },
    "papermill": {
     "duration": 18.640054,
     "end_time": "2021-03-19T13:31:56.069923",
     "exception": false,
     "start_time": "2021-03-19T13:31:37.429869",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "predicting...\n",
      "118/118 [==============================] - 14s 121ms/step\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "   No Defect       0.88      0.87      0.87      1771\n",
      "      Defect       0.89      0.89      0.89      2000\n",
      "\n",
      "    accuracy                           0.88      3771\n",
      "   macro avg       0.88      0.88      0.88      3771\n",
      "weighted avg       0.88      0.88      0.88      3771\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Save results as CSV files\n",
    "model.load_weights('model.h5')\n",
    "#saved_model = load_model(checkpoint_best_model)\n",
    "\n",
    "category_names = [\"No Defect\", \"Defect\"]\n",
    "\n",
    "y_test = df_test['defect_label']\n",
    "\n",
    "STEP_SIZE_TEST = compute_steps_per_epoch(y_test.shape[0])\n",
    "print('predicting...')\n",
    "\n",
    "predictions = model.predict(\n",
    "    test_gen,\n",
    "    steps = STEP_SIZE_TEST,\n",
    "    verbose=1)\n",
    "\n",
    "y_pred = (predictions > 0.5).astype(\"int32\")\n",
    "\n",
    "# classification report\n",
    "report_string = classification_report(y_test, y_pred, target_names=category_names)\n",
    "print(report_string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T13:32:03.451312Z",
     "iopub.status.busy": "2021-03-19T13:32:03.450413Z",
     "iopub.status.idle": "2021-03-19T13:32:03.646499Z",
     "shell.execute_reply": "2021-03-19T13:32:03.647072Z"
    },
    "papermill": {
     "duration": 3.837162,
     "end_time": "2021-03-19T13:32:03.647209",
     "exception": false,
     "start_time": "2021-03-19T13:31:59.810047",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_cm(y_true, y_pred, figsize=(6,5)):\n",
    "    cm = confusion_matrix(y_true, y_pred, labels=np.unique(y_true))\n",
    "    cm_sum = np.sum(cm, axis=1, keepdims=True)\n",
    "    cm_perc = cm / cm_sum.astype(float) * 100\n",
    "    annot = np.empty_like(cm).astype(str)\n",
    "    nrows, ncols = cm.shape\n",
    "    for i in range(nrows):\n",
    "        for j in range(ncols):\n",
    "            c = cm[i, j]\n",
    "            p = cm_perc[i, j]\n",
    "            if i == j:\n",
    "                s = cm_sum[i]\n",
    "                annot[i, j] = '%.1f%%\\n%d/%d' % (p, c, s)\n",
    "            elif c == 0:\n",
    "                annot[i, j] = ''\n",
    "            elif i<j:\n",
    "                annot[i, j] = '%.1f%%\\n%d\\nFP' % (p, c)\n",
    "            else:\n",
    "                annot[i, j] = '%.1f%%\\n%d\\nFN' % (p, c)\n",
    "    #cm = pd.DataFrame(cm, index=np.unique(y_true), columns=np.unique(y_true))\n",
    "    cm = pd.DataFrame(cm, index=category_names, columns=category_names)\n",
    "    cm.index.name = 'Actual'\n",
    "    cm.columns.name = 'Predicted'\n",
    "    fig, ax = plt.subplots(figsize=figsize)\n",
    "    sns.set(font_scale=1.4)\n",
    "    sns.heatmap(cm, cmap= \"YlGnBu\", annot=annot, fmt='', ax=ax, annot_kws={\"fontsize\":12})\n",
    "\n",
    "plot_cm(y_test, y_pred, figsize=(6,5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T13:32:10.828799Z",
     "iopub.status.busy": "2021-03-19T13:32:10.827846Z",
     "iopub.status.idle": "2021-03-19T13:32:11.208862Z",
     "shell.execute_reply": "2021-03-19T13:32:11.209253Z"
    },
    "papermill": {
     "duration": 4.034902,
     "end_time": "2021-03-19T13:32:11.209389",
     "exception": false,
     "start_time": "2021-03-19T13:32:07.174487",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC AUC=0.950\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve, auc, roc_auc_score\n",
    "\n",
    "# calculate scores\n",
    "auc = roc_auc_score(y_test, predictions)\n",
    "# summarize scores\n",
    "print('ROC AUC=%.3f' % (auc))\n",
    "# calculate roc curves\n",
    "\n",
    "ns_fpr, ns_tpr, _ = roc_curve(y_test, predictions)\n",
    "\n",
    "# plot the roc curve for the model\n",
    "plt.plot(ns_fpr, ns_tpr, linestyle='--', label=\"AUC = \"+ str(auc))\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
    "# axis labels\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "# show the legend\n",
    "plt.legend()\n",
    "# show the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-03-19T13:32:18.350203Z",
     "iopub.status.busy": "2021-03-19T13:32:18.349307Z",
     "iopub.status.idle": "2021-03-19T13:32:18.577095Z",
     "shell.execute_reply": "2021-03-19T13:32:18.576580Z"
    },
    "papermill": {
     "duration": 3.764023,
     "end_time": "2021-03-19T13:32:18.577196",
     "exception": false,
     "start_time": "2021-03-19T13:32:14.813173",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "model.save('severstal-classification-VGG.h5', save_format=\"h5\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 3.513549,
     "end_time": "2021-03-19T13:32:25.830961",
     "exception": false,
     "start_time": "2021-03-19T13:32:22.317412",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.8.3"
  },
  "papermill": {
   "duration": 5057.220036,
   "end_time": "2021-03-19T13:32:30.943316",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-03-19T12:08:13.723280",
   "version": "2.1.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}