Datasets
Models
Downloads: 1
1258 lines (1257 with data), 247.3 kB
{
"cells": [
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import pickle\n",
"import random\n",
"import glob\n",
"import datetime\n",
"import pandas as pd\n",
"import numpy as np\n",
"import cv2\n",
"import pydicom\n",
"from tqdm import tqdm\n",
"from joblib import delayed, Parallel\n",
"import zipfile\n",
"from pydicom.filebase import DicomBytesIO\n",
"import sys\n",
"from PIL import Image\n",
"import cv2\n",
"#from focal_loss import sparse_categorical_focal_loss\n",
"import keras\n",
"#import tensorflow_addons as tfa\n",
"from keras.preprocessing.image import ImageDataGenerator\n",
"from keras.models import model_from_json\n",
"import tensorflow as tf\n",
"import keras\n",
"from keras.models import Sequential, Model\n",
"from keras.layers import Dense, Conv2D, Flatten, MaxPooling2D, GlobalAveragePooling2D, Dropout\n",
"from keras.applications.inception_v3 import InceptionV3\n",
"\n",
"# importing pyplot and image from matplotlib \n",
"import matplotlib.pyplot as plt \n",
"import matplotlib.image as mpimg \n",
"from keras.optimizers import SGD\n",
"from keras import backend\n",
"\n",
"from keras.preprocessing import image\n",
"import albumentations as A\n",
"\n",
"\n",
"from sklearn.metrics import classification_report, confusion_matrix, roc_auc_score, roc_curve\n"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
"base_url = '/home/ubuntu/kaggle/rsna-intracranial-hemorrhage-detection/'\n",
"TRAIN_DIR = '/home/ubuntu/kaggle/rsna-intracranial-hemorrhage-detection/stage_2_train/'\n",
"TEST_DIR = '/home/ubuntu/kaggle/rsna-intracranial-hemorrhage-detection/stage_2_test/'\n",
"image_dir = '/home/ubuntu/kaggle/rsna-intracranial-hemorrhage-detection/png/train/adjacent-brain-cropped/'\n",
"save_dir = 'home/ubuntu/kaggle/models/'\n",
"os.listdir(base_url)\n",
"\n",
"def png(image): \n",
" return image + '.png'"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# focus_loss"
]
},
{
"cell_type": "code",
"execution_count": 76,
"metadata": {},
"outputs": [],
"source": [
"\n",
"#model.compile(loss=\"binary_crossentropy\", optimizer=keras.optimizers.Adam(), metrics=[weighted_loss])\n",
"#model.compile(optimizer=SGD(lr=learning_rate, momentum=0.9), loss=[focal_loss(6)], metrics=['accuracy'])\n",
"#model.compile('sgd', loss=tf.keras.losses.SigmoidFocalCrossEntropy())"
]
},
{
"cell_type": "code",
"execution_count": 73,
"metadata": {},
"outputs": [],
"source": [
"\n",
"decay_steps = 1000\n",
"initial_learning_rate = 1e-2\n",
"lr_decayed_fn = tf.keras.experimental.CosineDecay(\n",
" initial_learning_rate, decay_steps)\n",
"\n",
"opt = tf.keras.optimizers.SGD(learning_rate=lr_decayed_fn)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# weighted_metrics"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [],
"source": [
"from keras import backend as K\n",
"def _normalized_weighted_average(arr, weights=None):\n",
" \"\"\"\n",
" A simple Keras implementation that mimics that of \n",
" numpy.average(), specifically for this competition\n",
" \"\"\"\n",
" \n",
" if weights is not None:\n",
" scl = K.sum(weights)\n",
" weights = K.expand_dims(weights, axis=1)\n",
" return K.sum(K.dot(arr, weights), axis=1) / scl\n",
" return K.mean(arr, axis=1)\n",
"\n",
"\n",
"def weighted_loss(y_true, y_pred):\n",
" \"\"\"\n",
" Will be used as the metric in model.compile()\n",
" ---------------------------------------------\n",
" \n",
" Similar to the custom loss function 'weighted_log_loss()' above\n",
" but with normalized weights, which should be very similar \n",
" to the official competition metric:\n",
" https://www.kaggle.com/kambarakun/lb-probe-weights-n-of-positives-scoring\n",
" and hence:\n",
" sklearn.metrics.log_loss with sample weights\n",
" \"\"\"\n",
" \n",
" class_weights = K.variable([2., 1., 1., 1., 1., 1.])\n",
" \n",
" eps = K.epsilon()\n",
" \n",
" y_pred = K.clip(y_pred, eps, 1.0-eps)\n",
"\n",
" loss = -( y_true * K.log( y_pred)\n",
" + (1.0 - y_true) * K.log(1.0 - y_pred))\n",
" \n",
" loss_samples = _normalized_weighted_average(loss, class_weights)\n",
" \n",
" return K.mean(loss_samples)\n",
"\n",
"def auc(y_true, y_pred):\n",
" auc = tf.metrics.auc(y_true, y_pred)[1]\n",
" K.get_session().run(tf.local_variables_initializer())\n",
" return auc"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Data Generator"
]
},
{
"cell_type": "code",
"execution_count": 109,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Found 600937 validated image filenames.\n",
"Found 151865 validated image filenames.\n"
]
}
],
"source": [
"train_idg = ImageDataGenerator(rotation_range=360,\n",
" horizontal_flip=True,\n",
" validation_split=0.15,\n",
" rescale=1./255)\n",
"valid_idg = ImageDataGenerator()\n",
"training_data = pd.read_csv(f'train_0.csv') \n",
"training_data['Image'] = training_data['Image'].apply(png)\n",
"\n",
"validation_data = pd.read_csv(f'valid_0.csv')\n",
"validation_data['Image'] = validation_data['Image'].apply(png)\n",
"\n",
"columns=['any','epidural','intraparenchymal','intraventricular', 'subarachnoid','subdural']\n",
"\n",
"train_data_generator = train_idg.flow_from_dataframe(training_data, directory = image_dir,\n",
" x_col = \"Image\", y_col = columns,batch_size=64,\n",
" class_mode=\"raw\", target_size=(224,224), shuffle = True)\n",
"valid_data_generator = valid_idg.flow_from_dataframe(validation_data, directory = image_dir,\n",
" x_col = \"Image\", y_col = columns,batch_size=64,\n",
" class_mode = \"raw\",target_size=(224,224), shuffle = False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Model"
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"__________________________________________________________________________________________________\n",
"Layer (type) Output Shape Param # Connected to \n",
"==================================================================================================\n",
"input_7 (InputLayer) (None, 224, 224, 3) 0 \n",
"__________________________________________________________________________________________________\n",
"conv2d_565 (Conv2D) (None, 111, 111, 32) 864 input_7[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_565 (BatchN (None, 111, 111, 32) 96 conv2d_565[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_565 (Activation) (None, 111, 111, 32) 0 batch_normalization_565[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_566 (Conv2D) (None, 109, 109, 32) 9216 activation_565[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_566 (BatchN (None, 109, 109, 32) 96 conv2d_566[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_566 (Activation) (None, 109, 109, 32) 0 batch_normalization_566[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_567 (Conv2D) (None, 109, 109, 64) 18432 activation_566[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_567 (BatchN (None, 109, 109, 64) 192 conv2d_567[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_567 (Activation) (None, 109, 109, 64) 0 batch_normalization_567[0][0] \n",
"__________________________________________________________________________________________________\n",
"max_pooling2d_25 (MaxPooling2D) (None, 54, 54, 64) 0 activation_567[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_568 (Conv2D) (None, 54, 54, 80) 5120 max_pooling2d_25[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_568 (BatchN (None, 54, 54, 80) 240 conv2d_568[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_568 (Activation) (None, 54, 54, 80) 0 batch_normalization_568[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_569 (Conv2D) (None, 52, 52, 192) 138240 activation_568[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_569 (BatchN (None, 52, 52, 192) 576 conv2d_569[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_569 (Activation) (None, 52, 52, 192) 0 batch_normalization_569[0][0] \n",
"__________________________________________________________________________________________________\n",
"max_pooling2d_26 (MaxPooling2D) (None, 25, 25, 192) 0 activation_569[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_573 (Conv2D) (None, 25, 25, 64) 12288 max_pooling2d_26[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_573 (BatchN (None, 25, 25, 64) 192 conv2d_573[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_573 (Activation) (None, 25, 25, 64) 0 batch_normalization_573[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_571 (Conv2D) (None, 25, 25, 48) 9216 max_pooling2d_26[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_574 (Conv2D) (None, 25, 25, 96) 55296 activation_573[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_571 (BatchN (None, 25, 25, 48) 144 conv2d_571[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_574 (BatchN (None, 25, 25, 96) 288 conv2d_574[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_571 (Activation) (None, 25, 25, 48) 0 batch_normalization_571[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_574 (Activation) (None, 25, 25, 96) 0 batch_normalization_574[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_55 (AveragePo (None, 25, 25, 192) 0 max_pooling2d_26[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_570 (Conv2D) (None, 25, 25, 64) 12288 max_pooling2d_26[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_572 (Conv2D) (None, 25, 25, 64) 76800 activation_571[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_575 (Conv2D) (None, 25, 25, 96) 82944 activation_574[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_576 (Conv2D) (None, 25, 25, 32) 6144 average_pooling2d_55[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_570 (BatchN (None, 25, 25, 64) 192 conv2d_570[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_572 (BatchN (None, 25, 25, 64) 192 conv2d_572[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_575 (BatchN (None, 25, 25, 96) 288 conv2d_575[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_576 (BatchN (None, 25, 25, 32) 96 conv2d_576[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_570 (Activation) (None, 25, 25, 64) 0 batch_normalization_570[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_572 (Activation) (None, 25, 25, 64) 0 batch_normalization_572[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_575 (Activation) (None, 25, 25, 96) 0 batch_normalization_575[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_576 (Activation) (None, 25, 25, 32) 0 batch_normalization_576[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed0 (Concatenate) (None, 25, 25, 256) 0 activation_570[0][0] \n",
" activation_572[0][0] \n",
" activation_575[0][0] \n",
" activation_576[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_580 (Conv2D) (None, 25, 25, 64) 16384 mixed0[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_580 (BatchN (None, 25, 25, 64) 192 conv2d_580[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_580 (Activation) (None, 25, 25, 64) 0 batch_normalization_580[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_578 (Conv2D) (None, 25, 25, 48) 12288 mixed0[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_581 (Conv2D) (None, 25, 25, 96) 55296 activation_580[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_578 (BatchN (None, 25, 25, 48) 144 conv2d_578[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_581 (BatchN (None, 25, 25, 96) 288 conv2d_581[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_578 (Activation) (None, 25, 25, 48) 0 batch_normalization_578[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_581 (Activation) (None, 25, 25, 96) 0 batch_normalization_581[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_56 (AveragePo (None, 25, 25, 256) 0 mixed0[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_577 (Conv2D) (None, 25, 25, 64) 16384 mixed0[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_579 (Conv2D) (None, 25, 25, 64) 76800 activation_578[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_582 (Conv2D) (None, 25, 25, 96) 82944 activation_581[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_583 (Conv2D) (None, 25, 25, 64) 16384 average_pooling2d_56[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_577 (BatchN (None, 25, 25, 64) 192 conv2d_577[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_579 (BatchN (None, 25, 25, 64) 192 conv2d_579[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_582 (BatchN (None, 25, 25, 96) 288 conv2d_582[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_583 (BatchN (None, 25, 25, 64) 192 conv2d_583[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_577 (Activation) (None, 25, 25, 64) 0 batch_normalization_577[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_579 (Activation) (None, 25, 25, 64) 0 batch_normalization_579[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_582 (Activation) (None, 25, 25, 96) 0 batch_normalization_582[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_583 (Activation) (None, 25, 25, 64) 0 batch_normalization_583[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed1 (Concatenate) (None, 25, 25, 288) 0 activation_577[0][0] \n",
" activation_579[0][0] \n",
" activation_582[0][0] \n",
" activation_583[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_587 (Conv2D) (None, 25, 25, 64) 18432 mixed1[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_587 (BatchN (None, 25, 25, 64) 192 conv2d_587[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_587 (Activation) (None, 25, 25, 64) 0 batch_normalization_587[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_585 (Conv2D) (None, 25, 25, 48) 13824 mixed1[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_588 (Conv2D) (None, 25, 25, 96) 55296 activation_587[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_585 (BatchN (None, 25, 25, 48) 144 conv2d_585[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_588 (BatchN (None, 25, 25, 96) 288 conv2d_588[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_585 (Activation) (None, 25, 25, 48) 0 batch_normalization_585[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_588 (Activation) (None, 25, 25, 96) 0 batch_normalization_588[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_57 (AveragePo (None, 25, 25, 288) 0 mixed1[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_584 (Conv2D) (None, 25, 25, 64) 18432 mixed1[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_586 (Conv2D) (None, 25, 25, 64) 76800 activation_585[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_589 (Conv2D) (None, 25, 25, 96) 82944 activation_588[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_590 (Conv2D) (None, 25, 25, 64) 18432 average_pooling2d_57[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_584 (BatchN (None, 25, 25, 64) 192 conv2d_584[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_586 (BatchN (None, 25, 25, 64) 192 conv2d_586[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_589 (BatchN (None, 25, 25, 96) 288 conv2d_589[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_590 (BatchN (None, 25, 25, 64) 192 conv2d_590[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_584 (Activation) (None, 25, 25, 64) 0 batch_normalization_584[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_586 (Activation) (None, 25, 25, 64) 0 batch_normalization_586[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_589 (Activation) (None, 25, 25, 96) 0 batch_normalization_589[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_590 (Activation) (None, 25, 25, 64) 0 batch_normalization_590[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed2 (Concatenate) (None, 25, 25, 288) 0 activation_584[0][0] \n",
" activation_586[0][0] \n",
" activation_589[0][0] \n",
" activation_590[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_592 (Conv2D) (None, 25, 25, 64) 18432 mixed2[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_592 (BatchN (None, 25, 25, 64) 192 conv2d_592[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_592 (Activation) (None, 25, 25, 64) 0 batch_normalization_592[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_593 (Conv2D) (None, 25, 25, 96) 55296 activation_592[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_593 (BatchN (None, 25, 25, 96) 288 conv2d_593[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_593 (Activation) (None, 25, 25, 96) 0 batch_normalization_593[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_591 (Conv2D) (None, 12, 12, 384) 995328 mixed2[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_594 (Conv2D) (None, 12, 12, 96) 82944 activation_593[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_591 (BatchN (None, 12, 12, 384) 1152 conv2d_591[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_594 (BatchN (None, 12, 12, 96) 288 conv2d_594[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_591 (Activation) (None, 12, 12, 384) 0 batch_normalization_591[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_594 (Activation) (None, 12, 12, 96) 0 batch_normalization_594[0][0] \n",
"__________________________________________________________________________________________________\n",
"max_pooling2d_27 (MaxPooling2D) (None, 12, 12, 288) 0 mixed2[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed3 (Concatenate) (None, 12, 12, 768) 0 activation_591[0][0] \n",
" activation_594[0][0] \n",
" max_pooling2d_27[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_599 (Conv2D) (None, 12, 12, 128) 98304 mixed3[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_599 (BatchN (None, 12, 12, 128) 384 conv2d_599[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_599 (Activation) (None, 12, 12, 128) 0 batch_normalization_599[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_600 (Conv2D) (None, 12, 12, 128) 114688 activation_599[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_600 (BatchN (None, 12, 12, 128) 384 conv2d_600[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_600 (Activation) (None, 12, 12, 128) 0 batch_normalization_600[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_596 (Conv2D) (None, 12, 12, 128) 98304 mixed3[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_601 (Conv2D) (None, 12, 12, 128) 114688 activation_600[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_596 (BatchN (None, 12, 12, 128) 384 conv2d_596[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_601 (BatchN (None, 12, 12, 128) 384 conv2d_601[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_596 (Activation) (None, 12, 12, 128) 0 batch_normalization_596[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_601 (Activation) (None, 12, 12, 128) 0 batch_normalization_601[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_597 (Conv2D) (None, 12, 12, 128) 114688 activation_596[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_602 (Conv2D) (None, 12, 12, 128) 114688 activation_601[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_597 (BatchN (None, 12, 12, 128) 384 conv2d_597[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_602 (BatchN (None, 12, 12, 128) 384 conv2d_602[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_597 (Activation) (None, 12, 12, 128) 0 batch_normalization_597[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_602 (Activation) (None, 12, 12, 128) 0 batch_normalization_602[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_58 (AveragePo (None, 12, 12, 768) 0 mixed3[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_595 (Conv2D) (None, 12, 12, 192) 147456 mixed3[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_598 (Conv2D) (None, 12, 12, 192) 172032 activation_597[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_603 (Conv2D) (None, 12, 12, 192) 172032 activation_602[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_604 (Conv2D) (None, 12, 12, 192) 147456 average_pooling2d_58[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_595 (BatchN (None, 12, 12, 192) 576 conv2d_595[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_598 (BatchN (None, 12, 12, 192) 576 conv2d_598[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_603 (BatchN (None, 12, 12, 192) 576 conv2d_603[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_604 (BatchN (None, 12, 12, 192) 576 conv2d_604[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_595 (Activation) (None, 12, 12, 192) 0 batch_normalization_595[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_598 (Activation) (None, 12, 12, 192) 0 batch_normalization_598[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_603 (Activation) (None, 12, 12, 192) 0 batch_normalization_603[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_604 (Activation) (None, 12, 12, 192) 0 batch_normalization_604[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed4 (Concatenate) (None, 12, 12, 768) 0 activation_595[0][0] \n",
" activation_598[0][0] \n",
" activation_603[0][0] \n",
" activation_604[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_609 (Conv2D) (None, 12, 12, 160) 122880 mixed4[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_609 (BatchN (None, 12, 12, 160) 480 conv2d_609[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_609 (Activation) (None, 12, 12, 160) 0 batch_normalization_609[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_610 (Conv2D) (None, 12, 12, 160) 179200 activation_609[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_610 (BatchN (None, 12, 12, 160) 480 conv2d_610[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_610 (Activation) (None, 12, 12, 160) 0 batch_normalization_610[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_606 (Conv2D) (None, 12, 12, 160) 122880 mixed4[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_611 (Conv2D) (None, 12, 12, 160) 179200 activation_610[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_606 (BatchN (None, 12, 12, 160) 480 conv2d_606[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_611 (BatchN (None, 12, 12, 160) 480 conv2d_611[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_606 (Activation) (None, 12, 12, 160) 0 batch_normalization_606[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_611 (Activation) (None, 12, 12, 160) 0 batch_normalization_611[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_607 (Conv2D) (None, 12, 12, 160) 179200 activation_606[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_612 (Conv2D) (None, 12, 12, 160) 179200 activation_611[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_607 (BatchN (None, 12, 12, 160) 480 conv2d_607[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_612 (BatchN (None, 12, 12, 160) 480 conv2d_612[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_607 (Activation) (None, 12, 12, 160) 0 batch_normalization_607[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_612 (Activation) (None, 12, 12, 160) 0 batch_normalization_612[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_59 (AveragePo (None, 12, 12, 768) 0 mixed4[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_605 (Conv2D) (None, 12, 12, 192) 147456 mixed4[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_608 (Conv2D) (None, 12, 12, 192) 215040 activation_607[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_613 (Conv2D) (None, 12, 12, 192) 215040 activation_612[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_614 (Conv2D) (None, 12, 12, 192) 147456 average_pooling2d_59[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_605 (BatchN (None, 12, 12, 192) 576 conv2d_605[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_608 (BatchN (None, 12, 12, 192) 576 conv2d_608[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_613 (BatchN (None, 12, 12, 192) 576 conv2d_613[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_614 (BatchN (None, 12, 12, 192) 576 conv2d_614[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_605 (Activation) (None, 12, 12, 192) 0 batch_normalization_605[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_608 (Activation) (None, 12, 12, 192) 0 batch_normalization_608[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_613 (Activation) (None, 12, 12, 192) 0 batch_normalization_613[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_614 (Activation) (None, 12, 12, 192) 0 batch_normalization_614[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed5 (Concatenate) (None, 12, 12, 768) 0 activation_605[0][0] \n",
" activation_608[0][0] \n",
" activation_613[0][0] \n",
" activation_614[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_619 (Conv2D) (None, 12, 12, 160) 122880 mixed5[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_619 (BatchN (None, 12, 12, 160) 480 conv2d_619[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_619 (Activation) (None, 12, 12, 160) 0 batch_normalization_619[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_620 (Conv2D) (None, 12, 12, 160) 179200 activation_619[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_620 (BatchN (None, 12, 12, 160) 480 conv2d_620[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_620 (Activation) (None, 12, 12, 160) 0 batch_normalization_620[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_616 (Conv2D) (None, 12, 12, 160) 122880 mixed5[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_621 (Conv2D) (None, 12, 12, 160) 179200 activation_620[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_616 (BatchN (None, 12, 12, 160) 480 conv2d_616[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_621 (BatchN (None, 12, 12, 160) 480 conv2d_621[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_616 (Activation) (None, 12, 12, 160) 0 batch_normalization_616[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_621 (Activation) (None, 12, 12, 160) 0 batch_normalization_621[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_617 (Conv2D) (None, 12, 12, 160) 179200 activation_616[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_622 (Conv2D) (None, 12, 12, 160) 179200 activation_621[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_617 (BatchN (None, 12, 12, 160) 480 conv2d_617[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_622 (BatchN (None, 12, 12, 160) 480 conv2d_622[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_617 (Activation) (None, 12, 12, 160) 0 batch_normalization_617[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_622 (Activation) (None, 12, 12, 160) 0 batch_normalization_622[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_60 (AveragePo (None, 12, 12, 768) 0 mixed5[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_615 (Conv2D) (None, 12, 12, 192) 147456 mixed5[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_618 (Conv2D) (None, 12, 12, 192) 215040 activation_617[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_623 (Conv2D) (None, 12, 12, 192) 215040 activation_622[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_624 (Conv2D) (None, 12, 12, 192) 147456 average_pooling2d_60[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_615 (BatchN (None, 12, 12, 192) 576 conv2d_615[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_618 (BatchN (None, 12, 12, 192) 576 conv2d_618[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_623 (BatchN (None, 12, 12, 192) 576 conv2d_623[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_624 (BatchN (None, 12, 12, 192) 576 conv2d_624[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_615 (Activation) (None, 12, 12, 192) 0 batch_normalization_615[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_618 (Activation) (None, 12, 12, 192) 0 batch_normalization_618[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_623 (Activation) (None, 12, 12, 192) 0 batch_normalization_623[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_624 (Activation) (None, 12, 12, 192) 0 batch_normalization_624[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed6 (Concatenate) (None, 12, 12, 768) 0 activation_615[0][0] \n",
" activation_618[0][0] \n",
" activation_623[0][0] \n",
" activation_624[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_629 (Conv2D) (None, 12, 12, 192) 147456 mixed6[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_629 (BatchN (None, 12, 12, 192) 576 conv2d_629[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_629 (Activation) (None, 12, 12, 192) 0 batch_normalization_629[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_630 (Conv2D) (None, 12, 12, 192) 258048 activation_629[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_630 (BatchN (None, 12, 12, 192) 576 conv2d_630[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_630 (Activation) (None, 12, 12, 192) 0 batch_normalization_630[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_626 (Conv2D) (None, 12, 12, 192) 147456 mixed6[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_631 (Conv2D) (None, 12, 12, 192) 258048 activation_630[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_626 (BatchN (None, 12, 12, 192) 576 conv2d_626[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_631 (BatchN (None, 12, 12, 192) 576 conv2d_631[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_626 (Activation) (None, 12, 12, 192) 0 batch_normalization_626[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_631 (Activation) (None, 12, 12, 192) 0 batch_normalization_631[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_627 (Conv2D) (None, 12, 12, 192) 258048 activation_626[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_632 (Conv2D) (None, 12, 12, 192) 258048 activation_631[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_627 (BatchN (None, 12, 12, 192) 576 conv2d_627[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_632 (BatchN (None, 12, 12, 192) 576 conv2d_632[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_627 (Activation) (None, 12, 12, 192) 0 batch_normalization_627[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_632 (Activation) (None, 12, 12, 192) 0 batch_normalization_632[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_61 (AveragePo (None, 12, 12, 768) 0 mixed6[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_625 (Conv2D) (None, 12, 12, 192) 147456 mixed6[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_628 (Conv2D) (None, 12, 12, 192) 258048 activation_627[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_633 (Conv2D) (None, 12, 12, 192) 258048 activation_632[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_634 (Conv2D) (None, 12, 12, 192) 147456 average_pooling2d_61[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_625 (BatchN (None, 12, 12, 192) 576 conv2d_625[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_628 (BatchN (None, 12, 12, 192) 576 conv2d_628[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_633 (BatchN (None, 12, 12, 192) 576 conv2d_633[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_634 (BatchN (None, 12, 12, 192) 576 conv2d_634[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_625 (Activation) (None, 12, 12, 192) 0 batch_normalization_625[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_628 (Activation) (None, 12, 12, 192) 0 batch_normalization_628[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_633 (Activation) (None, 12, 12, 192) 0 batch_normalization_633[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_634 (Activation) (None, 12, 12, 192) 0 batch_normalization_634[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed7 (Concatenate) (None, 12, 12, 768) 0 activation_625[0][0] \n",
" activation_628[0][0] \n",
" activation_633[0][0] \n",
" activation_634[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_637 (Conv2D) (None, 12, 12, 192) 147456 mixed7[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_637 (BatchN (None, 12, 12, 192) 576 conv2d_637[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_637 (Activation) (None, 12, 12, 192) 0 batch_normalization_637[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_638 (Conv2D) (None, 12, 12, 192) 258048 activation_637[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_638 (BatchN (None, 12, 12, 192) 576 conv2d_638[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_638 (Activation) (None, 12, 12, 192) 0 batch_normalization_638[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_635 (Conv2D) (None, 12, 12, 192) 147456 mixed7[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_639 (Conv2D) (None, 12, 12, 192) 258048 activation_638[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_635 (BatchN (None, 12, 12, 192) 576 conv2d_635[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_639 (BatchN (None, 12, 12, 192) 576 conv2d_639[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_635 (Activation) (None, 12, 12, 192) 0 batch_normalization_635[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_639 (Activation) (None, 12, 12, 192) 0 batch_normalization_639[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_636 (Conv2D) (None, 5, 5, 320) 552960 activation_635[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_640 (Conv2D) (None, 5, 5, 192) 331776 activation_639[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_636 (BatchN (None, 5, 5, 320) 960 conv2d_636[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_640 (BatchN (None, 5, 5, 192) 576 conv2d_640[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_636 (Activation) (None, 5, 5, 320) 0 batch_normalization_636[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_640 (Activation) (None, 5, 5, 192) 0 batch_normalization_640[0][0] \n",
"__________________________________________________________________________________________________\n",
"max_pooling2d_28 (MaxPooling2D) (None, 5, 5, 768) 0 mixed7[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed8 (Concatenate) (None, 5, 5, 1280) 0 activation_636[0][0] \n",
" activation_640[0][0] \n",
" max_pooling2d_28[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_645 (Conv2D) (None, 5, 5, 448) 573440 mixed8[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_645 (BatchN (None, 5, 5, 448) 1344 conv2d_645[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_645 (Activation) (None, 5, 5, 448) 0 batch_normalization_645[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_642 (Conv2D) (None, 5, 5, 384) 491520 mixed8[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_646 (Conv2D) (None, 5, 5, 384) 1548288 activation_645[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_642 (BatchN (None, 5, 5, 384) 1152 conv2d_642[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_646 (BatchN (None, 5, 5, 384) 1152 conv2d_646[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_642 (Activation) (None, 5, 5, 384) 0 batch_normalization_642[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_646 (Activation) (None, 5, 5, 384) 0 batch_normalization_646[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_643 (Conv2D) (None, 5, 5, 384) 442368 activation_642[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_644 (Conv2D) (None, 5, 5, 384) 442368 activation_642[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_647 (Conv2D) (None, 5, 5, 384) 442368 activation_646[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_648 (Conv2D) (None, 5, 5, 384) 442368 activation_646[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_62 (AveragePo (None, 5, 5, 1280) 0 mixed8[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_641 (Conv2D) (None, 5, 5, 320) 409600 mixed8[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_643 (BatchN (None, 5, 5, 384) 1152 conv2d_643[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_644 (BatchN (None, 5, 5, 384) 1152 conv2d_644[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_647 (BatchN (None, 5, 5, 384) 1152 conv2d_647[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_648 (BatchN (None, 5, 5, 384) 1152 conv2d_648[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_649 (Conv2D) (None, 5, 5, 192) 245760 average_pooling2d_62[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_641 (BatchN (None, 5, 5, 320) 960 conv2d_641[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_643 (Activation) (None, 5, 5, 384) 0 batch_normalization_643[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_644 (Activation) (None, 5, 5, 384) 0 batch_normalization_644[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_647 (Activation) (None, 5, 5, 384) 0 batch_normalization_647[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_648 (Activation) (None, 5, 5, 384) 0 batch_normalization_648[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_649 (BatchN (None, 5, 5, 192) 576 conv2d_649[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_641 (Activation) (None, 5, 5, 320) 0 batch_normalization_641[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed9_0 (Concatenate) (None, 5, 5, 768) 0 activation_643[0][0] \n",
" activation_644[0][0] \n",
"__________________________________________________________________________________________________\n",
"concatenate_13 (Concatenate) (None, 5, 5, 768) 0 activation_647[0][0] \n",
" activation_648[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_649 (Activation) (None, 5, 5, 192) 0 batch_normalization_649[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed9 (Concatenate) (None, 5, 5, 2048) 0 activation_641[0][0] \n",
" mixed9_0[0][0] \n",
" concatenate_13[0][0] \n",
" activation_649[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_654 (Conv2D) (None, 5, 5, 448) 917504 mixed9[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_654 (BatchN (None, 5, 5, 448) 1344 conv2d_654[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_654 (Activation) (None, 5, 5, 448) 0 batch_normalization_654[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_651 (Conv2D) (None, 5, 5, 384) 786432 mixed9[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_655 (Conv2D) (None, 5, 5, 384) 1548288 activation_654[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_651 (BatchN (None, 5, 5, 384) 1152 conv2d_651[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_655 (BatchN (None, 5, 5, 384) 1152 conv2d_655[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_651 (Activation) (None, 5, 5, 384) 0 batch_normalization_651[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_655 (Activation) (None, 5, 5, 384) 0 batch_normalization_655[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_652 (Conv2D) (None, 5, 5, 384) 442368 activation_651[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_653 (Conv2D) (None, 5, 5, 384) 442368 activation_651[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_656 (Conv2D) (None, 5, 5, 384) 442368 activation_655[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_657 (Conv2D) (None, 5, 5, 384) 442368 activation_655[0][0] \n",
"__________________________________________________________________________________________________\n",
"average_pooling2d_63 (AveragePo (None, 5, 5, 2048) 0 mixed9[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_650 (Conv2D) (None, 5, 5, 320) 655360 mixed9[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_652 (BatchN (None, 5, 5, 384) 1152 conv2d_652[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_653 (BatchN (None, 5, 5, 384) 1152 conv2d_653[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_656 (BatchN (None, 5, 5, 384) 1152 conv2d_656[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_657 (BatchN (None, 5, 5, 384) 1152 conv2d_657[0][0] \n",
"__________________________________________________________________________________________________\n",
"conv2d_658 (Conv2D) (None, 5, 5, 192) 393216 average_pooling2d_63[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_650 (BatchN (None, 5, 5, 320) 960 conv2d_650[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_652 (Activation) (None, 5, 5, 384) 0 batch_normalization_652[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_653 (Activation) (None, 5, 5, 384) 0 batch_normalization_653[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_656 (Activation) (None, 5, 5, 384) 0 batch_normalization_656[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_657 (Activation) (None, 5, 5, 384) 0 batch_normalization_657[0][0] \n",
"__________________________________________________________________________________________________\n",
"batch_normalization_658 (BatchN (None, 5, 5, 192) 576 conv2d_658[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_650 (Activation) (None, 5, 5, 320) 0 batch_normalization_650[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed9_1 (Concatenate) (None, 5, 5, 768) 0 activation_652[0][0] \n",
" activation_653[0][0] \n",
"__________________________________________________________________________________________________\n",
"concatenate_14 (Concatenate) (None, 5, 5, 768) 0 activation_656[0][0] \n",
" activation_657[0][0] \n",
"__________________________________________________________________________________________________\n",
"activation_658 (Activation) (None, 5, 5, 192) 0 batch_normalization_658[0][0] \n",
"__________________________________________________________________________________________________\n",
"mixed10 (Concatenate) (None, 5, 5, 2048) 0 activation_650[0][0] \n",
" mixed9_1[0][0] \n",
" concatenate_14[0][0] \n",
" activation_658[0][0] \n",
"__________________________________________________________________________________________________\n",
"global_average_pooling2d_7 (Glo (None, 2048) 0 mixed10[0][0] \n",
"__________________________________________________________________________________________________\n",
"dense_11 (Dense) (None, 256) 524544 global_average_pooling2d_7[0][0] \n",
"__________________________________________________________________________________________________\n",
"dense_12 (Dense) (None, 6) 1542 dense_11[0][0] \n",
"==================================================================================================\n",
"Total params: 22,328,870\n",
"Trainable params: 526,086\n",
"Non-trainable params: 21,802,784\n",
"__________________________________________________________________________________________________\n"
]
}
],
"source": [
"from keras.applications.inception_v3 import InceptionV3\n",
"\n",
"from keras.models import Model\n",
"from keras.layers import Dense, GlobalAveragePooling2D\n",
"\n",
"METRICS = [\n",
" tf.keras.metrics.TruePositives(name='tp'),\n",
" tf.keras.metrics.FalsePositives(name='fp'),\n",
" tf.keras.metrics.TrueNegatives(name='tn'),\n",
" tf.keras.metrics.FalseNegatives(name='fn'), \n",
" tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
" tf.keras.metrics.Precision(name='precision'),\n",
" tf.keras.metrics.Recall(name='recall'),\n",
" tf.keras.metrics.AUC(name='auc'),\n",
"]\n",
"\n",
"\n",
"\n",
"\n",
"\n",
"\n",
"\n",
"# create the base pre-trained model\n",
"base_model = InceptionV3(weights='imagenet', include_top=False, input_shape=(224,224,3))\n",
"\n",
"# add a global spatial average pooling layer\n",
"x = base_model.output\n",
"x = GlobalAveragePooling2D()(x)\n",
"# let's add a fully-connected layer\n",
"x = Dense(256, activation='relu')(x)\n",
"# and a logistic layer -- let's say we have 200 classes\n",
"\n",
"#initializer = keras.initializers.GlorotUniform()\n",
"#layer = tf.keras.layers.Dense(3, kernel_initializer=initializer)\n",
"\n",
"predictions = Dense(6, activation='sigmoid')(x)\n",
"#activation='sigmoid',kernel_initializer=keras.initializers.GlorotNormal()\n",
"# this is the model we will train\n",
"model = Model(inputs=base_model.input, outputs=predictions)\n",
"\n",
"# first: train only the top layers (which were randomly initialized)\n",
"# i.e. freeze all convolutional InceptionV3 layers\n",
"for layer in base_model.layers:\n",
" layer.trainable = False\n",
"\n",
"# compile the model (should be done *after* setting layers to non-trainable)\n",
"model.compile(loss=\"binary_crossentropy\", optimizer=keras.optimizers.Adam(), metrics=[weighted_loss,auc,'accuracy'])\n",
"#model.compile(loss=loss_func,\n",
"# optimizer=opt,\n",
"# metrics=METRICS)\n",
"model.summary()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Callbacks"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {},
"outputs": [],
"source": [
"batch_size=32\n",
"learning_rate=5e-4\n",
"num_epochs=5\n",
"decay_rate=0.8\n",
"decay_steps=1\n",
"reduce_lr = keras.callbacks.ReduceLROnPlateau(monitor='val_loss',\n",
" factor=0.5,\n",
" patience=2,\n",
" min_lr=1e-8,\n",
" mode=\"min\")\n",
"\n",
"scheduler = keras.callbacks.LearningRateScheduler(lambda epoch: learning_rate * pow(decay_rate, floor(num_epochs / decay_steps)))\n",
"Checkpoint= keras.callbacks.ModelCheckpoint(f\"Enet_model.h5\", monitor='val_accuracy', verbose=1, save_best_only=True,\n",
" save_weights_only=True,mode='max')\n",
"callback_list=[Checkpoint]\n",
"\n",
"# This saves the best model\n",
"#checkpoint = tf.keras.callbacks.ModelCheckpoint(save_dir+'InceptionResNetV2_0', \n",
"# monitor='val_accuracy', verbose=1, \n",
"# save_best_only=True, mode='max')\n",
"#callbacks_list = [checkpoint]\n",
"\n",
"\n",
"\n",
"# LOAD BEST MODEL to evaluate the performance of the model\n",
"#model.load_weights(\"/saved_models/model_\"+str(fold_var)+\".h5\")\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Model Fit"
]
},
{
"cell_type": "code",
"execution_count": 36,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/3\n",
"600/600 [==============================] - 740s 1s/step - loss: 0.1567 - weighted_loss: 0.1788 - auc: 0.8486 - acc: 0.9472 - val_loss: 0.1912 - val_weighted_loss: 0.2191 - val_auc: 0.8520 - val_acc: 0.9430\n",
"Epoch 2/3\n",
" 2/600 [..............................] - ETA: 1:00 - loss: 0.2008 - weighted_loss: 0.2268 - auc: 0.8488 - acc: 0.9284"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.6/site-packages/keras/callbacks.py:434: RuntimeWarning: Can save best model only with val_accuracy available, skipping.\n",
" 'skipping.' % (self.monitor), RuntimeWarning)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"600/600 [==============================] - 723s 1s/step - loss: 0.1562 - weighted_loss: 0.1778 - auc: 0.8540 - acc: 0.9469 - val_loss: 0.1716 - val_weighted_loss: 0.1962 - val_auc: 0.8574 - val_acc: 0.9460\n",
"Epoch 3/3\n",
"600/600 [==============================] - 722s 1s/step - loss: 0.1537 - weighted_loss: 0.1751 - auc: 0.8599 - acc: 0.9473 - val_loss: 0.1713 - val_weighted_loss: 0.1978 - val_auc: 0.8619 - val_acc: 0.9428\n"
]
}
],
"source": [
"\n",
"\n",
"\n",
"num_epochs = 3\n",
"\n",
"batch_size = 1000\n",
"training_steps = len(training_data) // batch_size\n",
"validation_step = len(validation_data) // batch_size\n",
"\n",
"\n",
"\n",
"\n",
"# FIT THE MODEL\n",
"history = model.fit_generator(train_data_generator,\n",
" epochs=num_epochs,steps_per_epoch=training_steps,\n",
" callbacks=callback_list,\n",
" validation_data=valid_data_generator,\n",
" validation_steps= validation_step) \n",
"\n",
"\n",
"\n",
"\n",
"\n",
"tf.keras.backend.clear_session()\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 63,
"metadata": {},
"outputs": [],
"source": [
"predict = model.evaluate_generator(valid_data_generator,steps=validation_step)"
]
},
{
"cell_type": "code",
"execution_count": 134,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"---------------\n",
"\n",
"validation data **loss** value = 0.16912455615026273\n",
"\n",
"---------------\n",
"\n",
"validation data **weighted_loss** value = 0.19489387107013867\n",
"\n",
"---------------\n",
"\n",
"validation data **AUC** value = 0.8610175129593602\n",
"\n",
"---------------\n",
"\n",
"validation data **accuracy** value = 0.8610175129593602\n",
"\n",
"---------------\n",
"\n"
]
}
],
"source": [
"print('\\n---------------\\n')\n",
"print('validation data **loss** value =', predict[0])\n",
"print('\\n---------------\\n')\n",
"print('validation data **weighted_loss** value = ', predict[1])\n",
"print('\\n---------------\\n')\n",
"print('validation data **AUC** value =',predict[2])\n",
"print('\\n---------------\\n')\n",
"print('validation data **accuracy** value =',predict[2])\n",
"print('\\n---------------\\n')"
]
},
{
"cell_type": "code",
"execution_count": 104,
"metadata": {},
"outputs": [],
"source": [
"def plot_training(H):\n",
" # construct a plot that plots and saves the training history\n",
" with plt.xkcd():\n",
" plt.figure(figsize = (10,10))\n",
" plt.plot(H.epoch,H.history[\"acc\"], label=\"train_acc\")\n",
" plt.plot(H.epoch,H.history[\"val_acc\"], label=\"val_acc\")\n",
" plt.title(\"Training Accuracy\")\n",
" plt.xlabel(\"Epoch #\")\n",
" plt.ylabel(\"Accuracy\")\n",
" plt.legend(loc=\"lower left\")\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 105,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x720 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_training(history)"
]
},
{
"cell_type": "code",
"execution_count": 106,
"metadata": {},
"outputs": [],
"source": [
"def plot_training(H):\n",
" # construct a plot that plots and saves the training history\n",
" with plt.xkcd():\n",
" plt.figure(figsize = (10,10))\n",
" plt.plot(H.epoch,H.history[\"loss\"], label=\"train_loss\")\n",
" plt.plot(H.epoch,H.history[\"val_loss\"], label=\"val_loss\")\n",
" plt.title(\"Training Loss\")\n",
" plt.xlabel(\"Epoch #\")\n",
" plt.ylabel(\"Loss\")\n",
" plt.legend(loc=\"lower left\")\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 107,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x720 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_training(history)"
]
},
{
"cell_type": "code",
"execution_count": 95,
"metadata": {},
"outputs": [],
"source": [
"def plot_metrics(history):\n",
" plt.figure(figsize = (15,15))\n",
" metrics = ['loss', 'auc']\n",
" for n, metric in enumerate(metrics):\n",
" \n",
" name = metric.replace(\"_\",\" \").capitalize()\n",
" plt.subplot(2,2,n+1)\n",
" \n",
" plt.plot(history.epoch, history.history[metric], color='red', label='Train')\n",
" plt.plot(history.epoch, history.history['val_'+metric],\n",
" color='orange', linestyle=\"--\", label='Val')\n",
" plt.xlabel('Epoch')\n",
" plt.ylabel(name)\n",
" if metric == 'loss':\n",
" plt.ylim([0, plt.ylim()[1]])\n",
" elif metric == 'auc':\n",
" plt.ylim([0.8,1])\n",
" else:\n",
" plt.ylim([0,1])\n",
"\n",
" plt.legend()"
]
},
{
"cell_type": "code",
"execution_count": 96,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 1080x1080 with 2 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plot_metrics(history)"
]
}
],
"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": "3.6.10"
}
},
"nbformat": 4,
"nbformat_minor": 4
}