def define(self, optimizer=Adam(lr=config.configured_learning_rate)):
"""Define the model layers
The defined model should be saved into base class
"""
self.optimizer = optimizer
#define the base model based on keras vgg16 model
keras_model = KerasVGG16(
weights=config.configured_pretrained_weight,
include_top=False, # if config.use_fineturn_model else True,
input_tensor=self.get_input_tensor(),
input_shape=self.get_input_shape(),
classes=len(config.configured_classes))
self.set_trainable(keras_model, False)
self.model_debug(keras_model)
x = keras_model.output
x = Flatten()(x)
x = Dense(4096, activation='elu', name='fc1')(x)
x = Dropout(0.6)(x)
x = Dense(self.noveltyDetectionLayerSize,
activation='elu',
name=self.noveltyDetectionLayerName)(x)
x = Dropout(0.6)(x)
predictions = Dense(len(config.configured_classes),
activation='softmax',
name='predictions')(x)
self.model = Model(input=keras_model.input, output=predictions)
"""
def _create(self):
base_model = KerasVGG16(weights='imagenet', include_top=False, input_tensor=self.get_input_tensor())
self.make_net_layers_non_trainable(base_model)
x = base_model.output
x = Flatten()(x)
x = Dense(4096, activation='elu', name='fc1')(x)
x = Dropout(0.6)(x)
x = Dense(self.noveltyDetectionLayerSize, activation='elu', name=self.noveltyDetectionLayerName)(x)
x = Dropout(0.6)(x)
predictions = Dense(len(config.classes), activation='softmax', name='predictions')(x)
self.model = Model(input=base_model.input, output=predictions)
def _create(self):
base_model = KerasVGG16(weights='imagenet',
include_top=False,
input_tensor=self.get_input_tensor())
#base_model = KerasVGG16(weights='imagenet', include_top=False, input_shape=(165, 100, 3))
self.make_net_layers_non_trainable(base_model)
x = base_model.output
print("the vgg16 layers is!!!!!!!!!!!!!!!!!!!!!!!")
print(base_model.layers[3].output.shape)
# print(base_model.layers[18].output.shape)
# print(base_model.layers[19].output.shape)
x = Flatten()(x)
x = Dense(4096, activation='elu', name='fc1')(x)
x = Dropout(0.6)(x)
feature = Dense(self.noveltyDetectionLayerSize,
activation='elu',
name=self.noveltyDetectionLayerName)(x)
x = Dropout(0.6)(feature)
predictions = Dense(len(config.classes),
activation='softmax',
name='predictions')(x)
# print("qqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqq")
# print(len(config.classes))
# self.model = Model(input=base_model.input, output=predictions).save()
if config.isCenterLoss:
print(config.isCenterLoss)
# input_2 chu
input_target = Input(shape=(None, ))
#print(input_target)
centers = Embedding(len(config.classes), 4096)(input_target)
print('center:', centers)
#print(centers.ndim)
# print(centers.shape)
# print(centers[:, 0])
# print(centers[:, 1])
center_loss = Lambda(
lambda x: K.sum(K.square(x[0] - x[1][:, 0]), 1, keepdims=True),
name='center_loss')([feature, centers])
self.center_model = Model(inputs=[base_model.input, input_target],
outputs=[predictions, center_loss])
elif config.isTripletLoss:
self.triplet_model = Model(input=base_model.input,
output=[predictions, feature])
else:
print(base_model.input) # tensor("input_1:0")
self.model = Model(input=base_model.input, output=predictions)
def _create(self):
# base_model = KerasVGG16(weights='imagenet', include_top=False, input_tensor=self.get_input_tensor())
# self.make_net_layers_non_trainable(base_model)
#
# test1 = base_model.get_layer('block5_conv3').output
# test1 = Lambda(lambda x: K.permute_dimensions(x, (0, 3, 1, 2)))(test1) #K.permute_dimensions(test1,(0, 3, 1, 2)) # shift to [Batch, Channel, Height, Width] ==> [B,512,28,28]
# test1 = Lambda(lambda x: K.reshape(x, [-1, 512, 14*14]))(test1)#K.reshape(test1, [-1, 512, 28 * 28])
# #test1_T = Lambda(lambda x: K.permute_dimensions(test1,[0,2,1]))(test1)
#
# x_value = Lambda(lambda x: K.batch_dot(x, K.permute_dimensions(x,[0,2,1])))(test1)
# x_value = Lambda(lambda x: K.reshape(x, [-1, 512*512]))(x_value) # K.reshape(x_value, [-1, 512 * 512])
# y_value = Lambda(lambda x: K.sqrt(x + 1e-10))(x_value) #K.sqrt(x_value + 1e-10)
# z_value = Lambda(lambda x: K.l2_normalize(x, axis=1))(y_value)#K.l2_normalize(y_value, axis=1)
#
# # x = base_model.output
# # z_value = Flatten()(z_value)
# x = Dense(128, activation='elu', name='fc1')(z_value)
# x = Dropout(0.6)(x)
# x = Dense(self.noveltyDetectionLayerSize, activation='elu', name=self.noveltyDetectionLayerName)(x)
# x = Dropout(0.6)(x)
# predictions = Dense(len(config.classes), activation='softmax', name='predictions')(x)
# self.model = Model(input=base_model.input, output=predictions)
#self.model.load_weights()
base_model = KerasVGG16(weights='imagenet',
include_top=False,
input_tensor=self.get_input_tensor())
self.make_net_layers_non_trainable(base_model)
x = base_model.output
print("the vgg16 layers is!!!!!!!!!!!!!!!!!!!!!!!")
print(base_model.layers[3].output.shape)
x = Flatten()(x)
x = Dense(4096, activation='elu', name='fc1')(x)
x = Dropout(0.6)(x)
x = Dense(self.noveltyDetectionLayerSize,
activation='elu',
name=self.noveltyDetectionLayerName)(x)
x = Dropout(0.6)(x)
predictions = Dense(len(config.classes),
activation='softmax',
name='predictions')(x)
self.model = Model(input=base_model.input, output=predictions)
self.model = load_model(
'C:/Users/Ctbri/Desktop/chenxingli/keras-transfer-learning-for-oxford102/trained/model-vgg16.h5'
)
def _create(self):
base_model = KerasVGG16(weights='imagenet', include_top=False, input_tensor=self.get_input_tensor())
self.make_net_layers_non_trainable(base_model)
x = base_model.output
x = Dropout(self.dropout)(x)
x = Flatten()(x)
#x = Dense(4096, activation='elu', name='fc1')(x)
#x = Dense(self.noveltyDetectionLayerSize, activation='elu', name=self.noveltyDetectionLayerName)(x)
#x = Dropout(0.6)(x)
if self.run_config.main.classification_type==_config.CLASSIFICATION_TYPE.CLASSIFICATION:
predictions = Dense(self.run_config.data.nb_classes, activation='softmax', name='predictions')(x)
elif self.run_config.main.classification_type==_config.CLASSIFICATION_TYPE.REGRESSION:
predictions = Dense(1, activation='linear', name='regression')(x)
elif self.run_config.main.classification_type==_config.CLASSIFICATION_TYPE.MULTIPLE_REGRESSION:
predictions = Dense(self.run_config.data.nb_classes, activation='linear', name='multiple_regression')(x)
else:
raise ValueError("Error, unknown classification_type: <%s>" % self.run_config.main.classification_type)
self.model = Model(input=base_model.input, output=predictions)