def gen_func(images):
return utils.balance(couples_generator(images))
train_couples_len = sum(1 for e in gen_func(images_train))
validation_couples_len = sum(1 for e in gen_func(images_validation))
test_couples_len = sum(1 for e in gen_func(images_test))
print('Training with %d couples.' % train_couples_len)
print('Validation with %d couples.' % validation_couples_len)
print('Testing with %d couples.' % test_couples_len)
print(input_shape)
data_gen = utils.get_ImageDataGenerator(all_images, input_shape)
train_iterator = utils.CouplesIterator(
utils.make_infinite(gen_func, images_train), input_shape, data_gen,
BATCH_SIZE)
validation_iterator = utils.CouplesIterator(
utils.make_infinite(gen_func, images_validation), input_shape, data_gen,
BATCH_SIZE)
test_iterator = utils.CouplesIterator(
utils.make_infinite(gen_func, images_test), input_shape, data_gen,
BATCH_SIZE)
model = network.create(input_shape, args.network)
network.compile(model, args.optimizer)
callbacks_list = [
ModelCheckpoint('best_pretrain_model.hdf5',
monitor='val_accuracy',
verbose=1,
print(c.shape)
return [f, c]
images_train = random.sample(all_image_names, int(len(all_image_names) * 0.9))
images_test = [i for i in all_image_names if i not in set(images_train)]
def couples_generator(images):
for i in images:
yield load_pair(i), labels[i]
def gen_func(images):
return couples_generator(images)
train_couples_len = sum(1 for e in gen_func(images_train))
test_couples_len = sum(1 for e in gen_func(images_test))
data_gen = utils.get_ImageDataGenerator(all_images, input_shape)
train_iterator = utils.CouplesIterator(utils.make_infinite(gen_func, images_train), input_shape, data_gen, BATCH_SIZE)
test_iterator = utils.CouplesIterator(utils.make_infinite(gen_func, images_test), input_shape, data_gen, BATCH_SIZE)
model = network.create(input_shape, args.network)
network.compile(model, args.optimizer)
model.fit_generator(train_iterator, steps_per_epoch=train_couples_len / BATCH_SIZE, epochs=EPOCHS)
score = model.evaluate_generator(test_iterator, steps=test_couples_len / BATCH_SIZE)
print(score)
random.shuffle(images)
for e in utils.balance(couples_generator(images)):
yield e
train_couples_len = sum(
1 for e in utils.balance(couples_generator(images_train)))
test_couples_len = sum(1
for e in utils.balance(couples_generator(images_test)))
print('Training with %d couples.' % train_couples_len)
print('Testing with %d couples.' % test_couples_len)
print(input_shape)
data_gen = utils.get_ImageDataGenerator(all_images, input_shape)
train_iterator = utils.CouplesIterator(inf_couples_generator(images_train),
input_shape, data_gen, BATCH_SIZE)
test_iterator = utils.CouplesIterator(inf_couples_generator(images_test),
input_shape, data_gen, BATCH_SIZE)
model = network.create(input_shape)
network.compile(model)
model.save('pretrain.h5')
model.fit_generator(train_iterator,
steps_per_epoch=train_couples_len / BATCH_SIZE,
epochs=EPOCHS)
model.save('pretrain.h5')
score = model.evaluate_generator(test_iterator,
