《Deep Learning wih Keras》CHAPTER03 notes: Deep Learning with ConvNets

基本概念

Local receptive fields
Shared weights and bias
Pooling layers

或者

An example of DCNN—LeNet

#define the convnet 
class LeNet:
	@staticmethod
	def build(input_shape, classes):
		model = Sequential()
		# CONV => RELU => POOL
		model.add(Conv2D(20, kernel_size=5, padding="same",
			input_shape=input_shape))
		model.add(Activation("relu"))
		model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))
		# CONV => RELU => POOL
		model.add(Conv2D(50, kernel_size=5, padding="same"))
		model.add(Activation("relu"))
		model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))
		# Flatten => RELU layers
		model.add(Flatten())
		model.add(Dense(500))
		model.add(Activation("relu"))
 
		# a softmax classifier
		model.add(Dense(classes))
		model.add(Activation("softmax"))

		return model

Recognizing CIFAR-10 images with deep learning

网络结构的定义

model = Sequential()
model.add(Conv2D(32, (3, 3), padding='same',
                 input_shape=(IMG_ROWS, IMG_COLS, IMG_CHANNELS)))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
  
model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(0.5))

model.add(Dense(NB_CLASSES))
model.add(Activation('softmax'))

model.summary()

网络结构和网络参数的保存

#save model
model_json = model.to_json()
open('cifar10_architecture.json', 'w').write(model_json)
model.save_weights('cifar10_weights.h5', overwrite=True)

网络结构和网络参数的加载恢复

#load model
model_architecture = 'cifar10_architecture.json'
model_weights = 'cifar10_weights.h5'
model = model_from_json(open(model_architecture).read())
model.load_weights(model_weights)

利用更深的结构提高网络性能

# network
model = Sequential()
 
model.add(Conv2D(32, kernel_size=3, padding='same',
                        input_shape=(IMG_ROWS, IMG_COLS, IMG_CHANNELS)))
model.add(Activation('relu'))
model.add(Conv2D(32, kernel_size=3, padding='same'))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
 
model.add(Conv2D(64, kernel_size=3, padding='same'))
model.add(Activation('relu'))
model.add(Conv2D(64, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
 
model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(0.5))

model.add(Dense(NB_CLASSES))
model.add(Activation('softmax'))

model.summary()