resnet50.py

import tensorflow as tf 
import os

from tensorflow.python.keras.applications import ResNet50
from tensorflow.python.keras.models import Sequential
from tensorflow.python.keras.layers import Dense, Flatten, GlobalAveragePooling2D

from tensorflow.python.keras.applications.resnet50 import preprocess_input
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator

#reset default graph
tf.reset_default_graph()

IMG_SIZE = 224
num_classes = 2
resnet_weight_paths = 'resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'
SAVE_MODEL = 'catvsdog_trainedmodel.hd5'

new_model = Sequential()

new_model.add(ResNet50(include_top=False,
				pooling='avg',
				weights=resnet_weight_paths))

new_model.add(Dense(num_classes,activation = 'softmax'))

#not to train first layer (ResNet) model. It is already trained

new_model.layers[0].trainable = False

#now compile the model
new_model.compile(optimizer='sgd',loss='categorical_crossentropy',metrics=['accuracy'])

#Fit model
data_generator = ImageDataGenerator()

train_generator = data_generator.flow_from_directory(
	'/media/madhi/347b8fa1-b2fa-499a-a0da-3dc2835341ac/madhi/Documents/python programs/intern/signzy/Assignment_2/training_set',
	target_size = (IMG_SIZE,IMG_SIZE),
	batch_size = 12,
	class_mode = 'categorical'
	)

validation_generator = data_generator.flow_from_directory(
	'/media/madhi/347b8fa1-b2fa-499a-a0da-3dc2835341ac/madhi/Documents/python programs/intern/signzy/Assignment_2/validation_set',
	target_size = (IMG_SIZE,IMG_SIZE),
	class_mode = 'categorical'
	)

#compile the model

new_model.fit_generator(
	train_generator,
	steps_per_epoch = 3,
	validation_data = validation_generator,
	validation_steps = 1
)

new_model.save(SAVE_MODEL)
print("Model is Saved..!")

#score trained model
scores = new_model.evaluate(validation_generator)

print("Test Loss:{}".format(scores[0]))
print("Test Accuracy:{}".format(scores[1]))