本文分享内容来自《深度学习计算机视觉实战》一书:
VGG 是由牛津大学 Visual Geometry Group 组在 2014 年提出的,算法的名字也由此而来。
VGG 的提出证明了网络深度的增加可以提升网络的性能。VGG 常用的网络结构有两种:VGG-16和 VGG-19,网络结构在本质上没有差异,只是网络的深度不一样。VGG 相比以前的网络有一个重要的改进,使用连续的小的卷积核代替较大的卷积核,如使用两个 3×3 的卷积核代替一个 5×5 的卷积核。这种替换不仅可以增加网络的深度,还可以显著地减少网络的计算量,对性能的提升有较大作用。
VGG 的网络结构如图 1.12 所示。图 1.12 中 D 列所示为 VGG-16,该网络包含 16 个隐藏层,其中 13 个卷积层和 3 个全连接层;E 列所示为 VGG-19,该网络包含 19 个隐藏层,其中 16 个卷的卷积都是采用 3×3 的卷积核,池化使用的是 kernel 大小为 2×2的最大池化操作。
图 1.12
VGG 网络是很多算法中特征提取的主干网络(backbone),具有重要的意义,但是 VGG 的显著缺点是参数量很大,需要耗费较大的计算资源,参数量很大部分来源于 3 个全连接层,所以将VGG 作为特征提取网络时,常用的做法是只使用网络的卷积部分。
VGG 网络的深入研究可以参考论文 Very Deep Convolutional Networks for Large-Scale Image Recognition。
VGG16的代码如下:
import inspect
import os
import numpy as np
import tensorflow as tf
import time
# 用于计算图像平均减法,以便数据标准化
VGG_MEAN = [103.939, 116.779, 123.68]
# VGG16
class Vgg16:
def __init__(self, vgg16_npy_path=None):
if vgg16_npy_path is None:
path = inspect.getfile(Vgg16)
path = os.path.abspath(os.path.join(path, os.pardir))
path = os.path.join(path, "vgg16.npy")
vgg16_npy_path = path
print(path)
self.data_dict = np.load(vgg16_npy_path, encoding='latin1').item()
print("npy file loaded")
# 网络搭建
def build(self, rgb):
"""
load variable from npy to build the VGG
:param rgb: rgb image [batch, height, width, 3] values scaled [0, 1]
"""
# 计算运行时间
start_time = time.time()
print("build model started")
rgb_scaled = rgb * 255.0
# 图像通道转换, RGB to BGR
red, green, blue = tf.split(axis=3, num_or_size_splits=3, value=rgb_scaled)
assert red.get_shape().as_list()[1:] == [224, 224, 1]
assert green.get_shape().as_list()[1:] == [224, 224, 1]
assert blue.get_shape().as_list()[1:] == [224, 224, 1]
bgr = tf.concat(axis=3, values=[
blue - VGG_MEAN[0],
green - VGG_MEAN[1],
red - VGG_MEAN[2],
])
assert bgr.get_shape().as_list()[1:] == [224, 224, 3]
# 搭建网络层,可以自己对照VGG16的网络结构学习
self.conv1_1 = self.conv_layer(bgr, "conv1_1")
self.conv1_2 = self.conv_layer(self.conv1_1, "conv1_2")
self.pool1 = self.max_pool(self.conv1_2, 'pool1')
self.conv2_1 = self.conv_layer(self.pool1, "conv2_1")
self.conv2_2 = self.conv_layer(self.conv2_1, "conv2_2")
self.pool2 = self.max_pool(self.conv2_2, 'pool2')
self.conv3_1 = self.conv_layer(self.pool2, "conv3_1")
self.conv3_2 = self.conv_layer(self.conv3_1, "conv3_2")
self.conv3_3 = self.conv_layer(self.conv3_2, "conv3_3")
self.pool3 = self.max_pool(self.conv3_3, 'pool3')
self.conv4_1 = self.conv_layer(self.pool3, "conv4_1")
self.conv4_2 = self.conv_layer(self.conv4_1, "conv4_2")
self.conv4_3 = self.conv_layer(self.conv4_2, "conv4_3")
self.pool4 = self.max_pool(self.conv4_3, 'pool4')
self.conv5_1 = self.conv_layer(self.pool4, "conv5_1")
self.conv5_2 = self.conv_layer(self.conv5_1, "conv5_2")
self.conv5_3 = self.conv_layer(self.conv5_2, "conv5_3")
self.pool5 = self.max_pool(self.conv5_3, 'pool5')
self.fc6 = self.fc_layer(self.pool5, "fc6")
assert self.fc6.get_shape().as_list()[1:] == [4096]
self.relu6 = tf.nn.relu(self.fc6)
# 全连接
self.fc7 = self.fc_layer(self.relu6, "fc7")
self.relu7 = tf.nn.relu(self.fc7)
self.fc8 = self.fc_layer(self.relu7, "fc8")
self.prob = tf.nn.softmax(self.fc8, name="prob")
self.data_dict = None
print(("build model finished: %ds" % (time.time() - start_time)))
# 平均池化
def avg_pool(self, bottom, name):
return tf.nn.avg_pool(bottom, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME', name=name)
# 最大池化
def max_pool(self, bottom, name):
return tf.nn.max_pool(bottom, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME', name=name)
# 卷积层
def conv_layer(self, bottom, name):
with tf.variable_scope(name):
filt = self.get_conv_filter(name)
conv = tf.nn.conv2d(bottom, filt, [1, 1, 1, 1], padding='SAME')
conv_biases = self.get_bias(name)
bias = tf.nn.bias_add(conv, conv_biases)
relu = tf.nn.relu(bias)
return relu
# 全连接
def fc_layer(self, bottom, name):
with tf.variable_scope(name):
shape = bottom.get_shape().as_list()
dim = 1
for d in shape[1:]:
dim *= d
x = tf.reshape(bottom, [-1, dim])
weights = self.get_fc_weight(name)
biases = self.get_bias(name)
# Fully connected layer. Note that the '+' operation automatically
# broadcasts the biases.
fc = tf.nn.bias_add(tf.matmul(x, weights), biases)
return fc
# 获取卷积核
def get_conv_filter(self, name):
return tf.constant(self.data_dict[name][0], name="filter")
# 获取偏置
def get_bias(self, name):
return tf.constant(self.data_dict[name][1], name="biases")
# 获取权重
def get_fc_weight(self, name):
return tf.constant(self.data_dict[name][0], name="weights")
VGG19的代码如下:
import os
import tensorflow as tf
import numpy as np
import time
import inspect
VGG_MEAN = [103.939, 116.779, 123.68]
# VGG19
class Vgg19:
def __init__(self, vgg19_npy_path=None):
if vgg19_npy_path is None:
path = inspect.getfile(Vgg19)
path = os.path.abspath(os.path.join(path, os.pardir))
path = os.path.join(path, "vgg19.npy")
vgg19_npy_path = path
print(vgg19_npy_path)
self.data_dict = np.load(vgg19_npy_path, encoding='latin1').item()
print("npy file loaded")
# 网络搭建
def build(self, rgb):
"""
load variable from npy to build the VGG
:param rgb: rgb image [batch, height, width, 3] values scaled [0, 1]
"""
start_time = time.time()
print("build model started")
rgb_scaled = rgb * 255.0
# Convert RGB to BGR
red, green, blue = tf.split(axis=3, num_or_size_splits=3, value=rgb_scaled)
assert red.get_shape().as_list()[1:] == [224, 224, 1]
assert green.get_shape().as_list()[1:] == [224, 224, 1]
assert blue.get_shape().as_list()[1:] == [224, 224, 1]
bgr = tf.concat(axis=3, values=[
blue - VGG_MEAN[0],
green - VGG_MEAN[1],
red - VGG_MEAN[2],
])
assert bgr.get_shape().as_list()[1:] == [224, 224, 3]
self.conv1_1 = self.conv_layer(bgr, "conv1_1")
self.conv1_2 = self.conv_layer(self.conv1_1, "conv1_2")
self.pool1 = self.max_pool(self.conv1_2, 'pool1')
self.conv2_1 = self.conv_layer(self.pool1, "conv2_1")
self.conv2_2 = self.conv_layer(self.conv2_1, "conv2_2")
self.pool2 = self.max_pool(self.conv2_2, 'pool2')
# VGG19和VGG16的网络层差异主要在后面部分,多了三层卷积
self.conv3_1 = self.conv_layer(self.pool2, "conv3_1")
self.conv3_2 = self.conv_layer(self.conv3_1, "conv3_2")
self.conv3_3 = self.conv_layer(self.conv3_2, "conv3_3")
self.conv3_4 = self.conv_layer(self.conv3_3, "conv3_4")
self.pool3 = self.max_pool(self.conv3_4, 'pool3')
self.conv4_1 = self.conv_layer(self.pool3, "conv4_1")
self.conv4_2 = self.conv_layer(self.conv4_1, "conv4_2")
self.conv4_3 = self.conv_layer(self.conv4_2, "conv4_3")
self.conv4_4 = self.conv_layer(self.conv4_3, "conv4_4")
self.pool4 = self.max_pool(self.conv4_4, 'pool4')
self.conv5_1 = self.conv_layer(self.pool4, "conv5_1")
self.conv5_2 = self.conv_layer(self.conv5_1, "conv5_2")
self.conv5_3 = self.conv_layer(self.conv5_2, "conv5_3")
self.conv5_4 = self.conv_layer(self.conv5_3, "conv5_4")
self.pool5 = self.max_pool(self.conv5_4, 'pool5')
self.fc6 = self.fc_layer(self.pool5, "fc6")
assert self.fc6.get_shape().as_list()[1:] == [4096]
self.relu6 = tf.nn.relu(self.fc6)
self.fc7 = self.fc_layer(self.relu6, "fc7")
self.relu7 = tf.nn.relu(self.fc7)
self.fc8 = self.fc_layer(self.relu7, "fc8")
self.prob = tf.nn.softmax(self.fc8, name="prob")
self.data_dict = None
print(("build model finished: %ds" % (time.time() - start_time)))
# 下述函数含义与VGG16相同
def avg_pool(self, bottom, name):
return tf.nn.avg_pool(bottom, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME', name=name)
def max_pool(self, bottom, name):
return tf.nn.max_pool(bottom, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME', name=name)
def conv_layer(self, bottom, name):
with tf.variable_scope(name):
filt = self.get_conv_filter(name)
conv = tf.nn.conv2d(bottom, filt, [1, 1, 1, 1], padding='SAME')
conv_biases = self.get_bias(name)
bias = tf.nn.bias_add(conv, conv_biases)
relu = tf.nn.relu(bias)
return relu
def fc_layer(self, bottom, name):
with tf.variable_scope(name):
shape = bottom.get_shape().as_list()
dim = 1
for d in shape[1:]:
dim *= d
x = tf.reshape(bottom, [-1, dim])
weights = self.get_fc_weight(name)
biases = self.get_bias(name)
# Fully connected layer. Note that the '+' operation automatically
# broadcasts the biases.
fc = tf.nn.bias_add(tf.matmul(x, weights), biases)
return fc
def get_conv_filter(self, name):
return tf.constant(self.data_dict[name][0], name="filter")
def get_bias(self, name):
return tf.constant(self.data_dict[name][1], name="biases")
def get_fc_weight(self, name):
return tf.constant(self.data_dict[name][0], name="weights")