def trainFromScratch_MobileNet():
from time import time
from keras.callbacks import TensorBoard
tensorboard = TensorBoard(log_dir="logs/resnet_CosmeticsBrands/{}".format(
time()),
histogram_freq=0,
write_graph=True,
write_images=True)
checkPoint = ModelCheckpoint(filepath='weights/ah_mobile_2cats.h5',
save_best_only=True)
model = MobileNet(weights="imagenet",
include_top=False,
input_shape=(image_size, image_size, 3))
genterator_training = ImageDataGenerator(
rescale=1. / 255,
#rotation_range=0.4,
#shear_range=0.2,
#zoom_range=0.1,
#width_shift_range=0.1,
#height_shift_range=0.1,
vertical_flip=True,
horizontal_flip=True)
augmented_training = genterator_training.flow_from_directory(
train_data,
target_size=(image_size, image_size),
class_mode="categorical",
shuffle=False,
batch_size=batch_size)
generator_test = ImageDataGenerator(rescale=1. / 255)
augmented_testing = generator_test.flow_from_directory(
test_data,
target_size=(image_size, image_size),
class_mode="categorical",
batch_size=batch_size,
shuffle=False)
# optimizer_1 = RMSprop(lr=0.0005)
# optimizer_2 = Adagrad()
# model.compile(optimizer_1, loss='categorical_crossentropy', metrics=["accuracy"])
training_bottleneck_features = model.predict_generator(
augmented_training, steps=training_instances // batch_size)
np.save(open('weights/ah_trainig_bottleneck.npy', 'w'),
training_bottleneck_features)
testing_bottleneck_features = model.predict_generator(
augmented_testing, steps=test_instances // batch_size)
np.save(open('weights/ah_testing_bottleneck.npy', 'w'),
testing_bottleneck_features)
def mobile(X_train, Y_train):
conv_base = MobileNet(weights='imagenet',
include_top=False,
input_shape=(128, 128, 3))
samples_generator = DataGenerator(X_train, Y_train, **params)
test_features = conv_base.predict_generator(samples_generator,
verbose=True)
return test_features
history_list = []
for i in range(len(Train_indices)):
print(i)
print('Done1')
# Need to get the correct samples for the training and val sets
X_train = [Trainval_ids[j] for j in Train_indices[i]]
Y_train = [Trainval_labels[j] for j in Train_indices[i]]
X_val = [Trainval_ids[j] for j in Val_indices[i]]
Y_val = [Trainval_labels[j] for j in Val_indices[i]]
print('Done2')
conv_base = MobileNet(weights='imagenet',
include_top=False,
input_shape=(128, 128, 3))
train_generator = DataGenerator(X_train, Y_train, **params)
X_train = conv_base.predict_generator(train_generator, verbose=True)
print('Done3')
val_generator = DataGenerator(X_val, Y_val, **params)
X_val = conv_base.predict_generator(val_generator, verbose=True)
Y_train = [Trainval_labels[j] for j in Train_indices[i]]
Y_val = [Trainval_labels[j] for j in Val_indices[i]]
# Need to have the same amount of labels as X_train and X_val, as theyre done in batches of 25:
rem1 = len(Y_train) % 25
if not rem1 == 0:
Y_train = Y_train[:-rem1][:]
rem2 = len(Y_val) % 25
if not rem2 == 0:
Y_val = Y_val[:-rem2][:]
print(X_train.shape)
