def pretrain_model():
model =Xception(include_top = False, weights = "imagenet",input_shape = (71,71,3))
x=model.output
x=GlobalAveragePooling2D()(x)
x=Dense(512,activation='relu')(x)
x = Dropout(0.8)(x)
x= Dense(256,activation='relu')(x)
x = Dropout(0.8)(x)
y=Dense(10,activation='softmax')(x)
model = Model(inputs = model.input, outputs =y)
for layer in model.layers[:30]:
layer.trainable = False
if layer.name.startswith('batch_normalization'):
layer.trainable = True
for layer in model.layers[30:]:
layer.trainable = True
#if layer.name.startswith('batch_normalization'):
# layer.trainable = True
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=["accuracy"])
return model
def get_kernel(params):
model = Xception(weights=None)
optimizer = SGD()
model.compile(
loss="sparse_categorical_crossentropy",
optimizer=optimizer,
metrics=["accuracy"],
)
return model
def create_model(self):
img_input_layer = tf.keras.layers.Input(self.img_input_dim,
name="img_input")
ann_input_layer = tf.keras.layers.Input(self.ann_input_dim,
name="ann_input")
## Uncomment lines according to model type
# If training Xception, use first line. If training InceptionResNetV2, use second line, etc.
model = Xception(input_tensor=img_input_layer,
include_top=False,
weights='imagenet')
# model = InceptionResNetV2(input_tensor=img_input_layer, include_top=False, weights='imagenet')
# model = EfficientNetB3(input_tensor=img_input_layer, include_top=False, weights='imagenet')
x = model(img_input_layer)
# output layers
x1 = tf.keras.layers.GlobalAveragePooling2D()(x)
x2 = tf.keras.layers.GlobalMaxPooling2D()(x)
x3 = tf.keras.layers.Flatten()(x)
# Annot model
embed = tf.keras.layers.Embedding(126964,
64,
input_length=self.ann_input_dim[0])
m = embed(ann_input_layer)
m = tf.keras.layers.Flatten()(m)
out = tf.keras.layers.Concatenate(axis=-1)([x1, x2, x3, m])
out = tf.keras.layers.Dropout(0.3)(out)
output_layer = tf.keras.layers.Dense(self.n_classes,
activation='softmax')(out)
model = tf.keras.models.Model(
inputs=[img_input_layer, ann_input_layer], outputs=output_layer)
model.compile(optimizer=tf.keras.optimizers.Adam(),
loss="categorical_crossentropy",
metrics=['acc'])
return model
def create_model(self):
input_layer = tf.keras.layers.Input(self.input_dim)
## Uncomment lines according to model type
## If training Xception, use first line. If training InceptionResNetV2, use second line, etc.
model = Xception(input_tensor=input_layer, include_top=False, weights='imagenet')
# model = InceptionResNetV2(input_tensor=input_layer, include_top=False, weights='imagenet')
x = model(input_layer)
# output layers
x1 = tf.keras.layers.GlobalAveragePooling2D()(x)
x2 = tf.keras.layers.GlobalMaxPooling2D()(x)
x3 = tf.keras.layers.Flatten()(x)
out = tf.keras.layers.Concatenate(axis=-1)([x1, x2, x3])
out = tf.keras.layers.Dropout(0.3)(out)
output_layer = tf.keras.layers.Dense(self.n_classes, activation='softmax')(out)
model = tf.keras.models.Model(inputs=input_layer, outputs=output_layer)
model.compile(optimizer=tf.keras.optimizers.Adam(),
loss="categorical_crossentropy", metrics=['acc'])
return model
dogs_model = InceptionV3(weights=None, classes=dogs_num_of_classes)
cats_model = InceptionV3(weights=None, classes=cats_num_of_classes)
elif model_name == 'inception_resnet_v2':
binary_model = InceptionResNetV2(weights=None, classes=num_of_classes)
dogs_model = InceptionResNetV2(weights=None, classes=dogs_num_of_classes)
cats_model = InceptionResNetV2(weights=None, classes=cats_num_of_classes)
elif model_name == 'xception':
binary_model = Xception(weights=None, classes=num_of_classes)
dogs_model = Xception(weights=None, classes=dogs_num_of_classes)
cats_model = Xception(weights=None, classes=cats_num_of_classes)
else:
raise ValueError(f"not supported model name {model_name}")
"""MODEL PARAMS AND CALLBACKS"""
# binary model #
binary_model.compile(optimizer='adam',
loss='BinaryCrossentropy',
metrics=['accuracy'])
binary_checkpoint = ModelCheckpoint(
filepath=f'advanced_weights_binary_{model_name}.hdf5',
save_best_only=True,
verbose=1)
binary_early_stopping = EarlyStopping(monitor='val_accuracy',
patience=10,
verbose=1)
binary_reduce_lr = ReduceLROnPlateau(patience=5, verbose=1)
# dogs model #
dogs_model.compile(optimizer='adam',
loss='categorical_crossentropy',
metrics=['accuracy'])
dogs_checkpoint = ModelCheckpoint(
checkpoint = ModelCheckpoint(filepath,
monitor='val_acc',
verbose=0,
save_best_only=True,
mode='max')
# Log the epoch detail into csv
csv_logger = CSVLogger(os.path.join(output_folder, modelname + '.csv'))
callbacks_list = [checkpoint, csv_logger]
# define model
model = Xception(include_top=True,
input_shape=(IMG_SIZE, IMG_SIZE, 3),
weights=None,
classes=4)
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.summary()
from tensorflow.keras.utils import plot_model
model_file = os.path.join(output_folder, modelname + "_model.png")
plot_model(model,
to_file=model_file,
show_shapes=True,
show_layer_names=False,
rankdir='TB')
# fit model
model.fit_generator(train_it,
validation_data=val_it,
epochs=50,
