经典CNN之：VGGNet

VGGNet（Visual Geometry Group）

VGGNet是牛津大学计算机视觉组和ＤｅｅｐＭｉｎｄ公司共同研发一种深度卷积网络，并且在２０１４年在ILSVRC比赛上获得了分类项目的第二名和定位项目的第一名，VggNet一共有六种不同的网络结构，但是每种结构都有含有５组卷积，每组卷积都使用３ｘ３的卷积核，每组卷积后进行一个２ｘ２最大池化，接下来是三个全连接层．在训练高级别的网络时，可以先训练低级别的网络，用前者获得的权重初始化高级别的网络，可以加速网络的收敛．
网络的几种不同配置

Ｃ级别的网络增加了１ｘ１的卷积核，它的作用主要是对像素进行线性变换，不改变输入出通道，和大小．
网络的参数

在有些卷积组中会出现连续的几个卷积层堆叠，使用３ｘ３的卷积核，这样的设计，既可以保证感受视野，有能减少卷基层的参数，　

例如：两个３ｘ３的卷积层叠加，等价于一个５ｘ５的卷积核的效果，３个３ｘ３的卷积核的叠加相当于一个７ｘ７的卷积核，而且参数更少．大约是7x7卷积核卷积层的（３＊３＊３)/(7*7)=0.55．而且拥有和７ｘ７卷积核一样的感受视野，三个卷积层的叠加，经过了更多次的非线性变换，对特征的学习能力更强．

两个３ｘ３的卷积核

ＩｍａｇｅＮｅｔ数据量很大，所以用ＣＩＦＡＲ１０来测试，cifar10数据的下载问题，主要实现最基本的Ｖｇｇｎｅｔ．由于cifar10图像很小，所以最后两个卷积组省去了池化操作．卷积核的数量也根据图像的大小进行了调整．

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import time
import cifar10_input     #用于加载和处理数据,来自tensorflow/models/tutotials/image/cifar10

网络参数

max_steps = ２０７００　#一共训练的两万多次，分了两次，中途保存过一次参数变量
batch_size = 256    # 小批量数据大小
s_times = 20　　　　　# 每轮训练数据的组数，每组为一batchsize

learning_rate = 0.0001
data_dir = 'cifar10data/cifar-10-batches-bin'    # 数据所在路径

权重，偏置初始化

# Xavier初始化方法
# 卷积权重(核）初始化
def init_conv_weights(shape, name):
    weights = tf.get_variable(name=name, shape=shape, dtype=tf.float32, 
                             initializer=tf.contrib.layers.xavier_initializer_conv2d())
    return weights

# 全连接权重初始化
def init_fc_weights(shape, name):
    weights = tf.get_variable(name=name, shape=shape, dtype=tf.float32,
                             initializer=tf.contrib.layers.xavier_initializer())
    return weights

# 偏置
def init_biases(shape, name):
    biases = tf.Variable(tf.random_normal(shape),name=name, dtype=tf.float32)
    return biases

卷积，池化，全连接

# 卷积
# 参数：输入张量,卷积核，偏置，卷积核在高和宽维度上移动的步长
def conv2d(input_tensor, weights, biases, s_h, s_w):
    conv = tf.nn.conv2d(input_tensor, weights, [1, s_h, s_w, 1], padding='SAME')
    return tf.nn.relu(conv + biases)

# 池化
# 参数：输入张量，池化核高和宽，池化核在高，宽维度上移动步长
def max_pool(input_tensor, k_h, k_w, s_h, s_w):
    return tf.nn.max_pool(input_tensor, ksize=[1, k_h, k_w, 1], strides=[1, s_h, s_w, 1], padding='SAME')

# 全链接
# 参数：输入张量，全连接权重，偏置
def fullc(input_tensor, weights, biases):
    return tf.nn.relu_layer(input_tensor, weights, biases)

占位节点

# 输入占位节点
images = tf.placeholder(tf.float32, [batch_size, 24 ,24 ,3])
labels = tf.placeholder(tf.int32, [batch_size])
# 正则
keep_prob = tf.placeholder(tf.float32)

构建网络

# 使用作用域对Ｏｐ进行封装，在使用tensorboard对网络结构进行可视化的效果较好

# 第一组卷积 conv3-16
with tf.name_scope('conv_group_1'):
    cw1 = init_conv_weights([3, 3, 3, 16], name='conv_w1')
    cb1 = init_biases([16], name='conv_b1')
    conv1 = conv2d(images, cw1, cb1, 1, 1)

# 最大池化 2x2
pool1 = max_pool(conv1, 2, 2, 2, 2)


# 第二组卷积　conv3-32
with tf.name_scope('conv_group_2'):
    cw2 = init_conv_weights([3, 3, 16, 32], name='conv_w2')
    cb2 = init_biases([32], name='conv_b2')
    conv2 = conv2d(pool1, cw2, cb2, 1, 1)

# 最大池化
pool2 = max_pool(conv2, 2, 2, 2, 2)

# 第三组卷积　conv3-64  conv3-64
with tf.name_scope('conv_group_3'):
    cw3 = init_conv_weights([3, 3, 32, 64], name='conv_w3')
    cb3 = init_biases([64], name='conv_b3')
    conv3 = conv2d(pool2, cw3, cb3, 1, 1)

    cw4 = init_conv_weights([3, 3, 64, 64], name='conv_w4')
    cb4 = init_biases([64], name='conv_b4')
    conv4 = conv2d(conv3, cw4, cb4, 1, 1)

# 最大池化
pool3 = max_pool(conv4, 2, 2, 2, 2)     

# 第四组卷积　conv3-128 conv3-128
with tf.name_scope('conv_group_4'):
    cw5 = init_conv_weights([3, 3, 64, 128], name='conv_w5')
    cb5 = init_biases([128], name='conv_b5')
    conv5 = conv2d(pool3, cw5, cb5, 1, 1)

    cw6 = init_conv_weights([3, 3, 128, 128], name='conv_w6')
    cb6 = init_biases([128], name='conv_b6')
    conv6 = conv2d(conv5, cw6, cb6, 1, 1)

# 此时张量的高和宽为　３ｘ３，继续池化为　２ｘ２
#pool4 = max_pool(conv6, 2, 2, 2, 2)

# 第五组卷积　conv3-256 conv3-256
with tf.name_scope('conv_group_5'):
    cw7 = init_conv_weights([3, 3, 128, 128], name='conv_w7')
    cb7 = init_biases([128], name='conv_b7')
    conv7 = conv2d(conv6, cw7, cb7, 1, 1)

    cw8 = init_conv_weights([3, 3, 128, 128], name='conv_w8')
    cb8 = init_biases([128], name='conv_b8')
    conv8 = conv2d(conv7, cw8, cb8, 1, 1)

# 此处应该还有一个池化，但是现在张量为３ｘ３，很小了，所以省略了池化

把最后一个卷积后的数据转换为一维

# 转换数据shape
reshape_conv8 = tf.reshape(conv8, [batch_size, -1])
　
n_in = reshape_conv8.get_shape()[-1].value

网络的全连接部分（两个隐藏层）

# 地一个全连接层命名空间
with tf.name_scope('fullc_1'):
    fw9 = init_fc_weights([n_in, 256], name='fullc_w9')
    fb9 = init_biases([256], name='fullc_b9')
    activation1 = fullc(reshape_conv8, fw9, fb9)

# dropout正则
drop_act1 = tf.nn.dropout(activation1, keep_prob)

with tf.name_scope('fullc_2'):
    fw10 = init_fc_weights([256, 256], name='fullc_w10')
    fb10 = init_biases([256], name='fullc_b10')
    activation2 = fullc(drop_act1, fw10, fb10)

# dropout正则
drop_act2 = tf.nn.dropout(activation2, keep_prob)

with tf.name_scope('fullc_3'):
    fw11 = init_fc_weights([256, 10], name='fullc_w11')
    fb11 = init_biases([10], name='full_b11')
    logits = tf.add(tf.matmul(drop_act2, fw11), fb11)
    output = tf.nn.softmax(logits)

损失函数优化器

cross_entropy= tf.nn.sparse_softmax_cross_entropy_with_logits( logits=logits, labels=labels)
cost = tf.reduce_mean(cross_entropy,name='Train_Cost')
optimizer = tf.train.AdamOptimizer(learning_rate).minimize(cost)

评估函数

# 用来评估测试数据的准确率
# 数据labels没有使用one-hot编码格式,labels是int32
def accuracy(labels, output):
    labels = tf.to_int64(labels)
    pred_result = tf.equal(labels, tf.argmax(output, 1))
    accu = tf.reduce_mean(tf.cast(pred_result, tf.float32))
    return accu

训练．测试数据

# 加载训练batch_size大小的数据，经过增强处理，剪裁，反转，等等
train_images, train_labels = cifar10_input.distorted_inputs(batch_size= batch_size, data_dir= data_dir)

Filling queue with 20000 CIFAR images before starting to train. This will take a few minutes.

# 加载测试数据，batch_size大小，不进行增强处理
test_images, test_labels = cifar10_input.inputs(batch_size= batch_size, data_dir= data_dir,eval_data= True)

Ｔｒａｉｎｉｎｇ

# Training
def training(sess, max_steps, s_times, keeprob, display):

    Cost = []
    for i in range(max_steps):
        for j in range(s_times):
            start = time.time()
            batch_images, batch_labels = sess.run([train_images, train_labels])
            opt = sess.run(optimizer, feed_dict={images:batch_images, labels:batch_labels,
                                                keep_prob:keeprob})
            every_batch_time = time.time() - start
        c = sess.run(cost, feed_dict={images:batch_images, labels:batch_labels,
                                        keep_prob:keeprob})

        Cost.append(c)
        if i % display == 0:
            samples_per_sec = float(batch_size) / every_batch_time
            format_str = 'Epoch %d: %d samples/sec, %.4f sec/batch, Cost : %.5f'
            print format_str%(i+display, samples_per_sec, every_batch_time, c)

    return Cost

会话

sess = tf.Session()
sess.run(tf.global_variables_initializer())
# 图片增强处理,时使用了16个线程加速,启动16个独立线程
tf.train.start_queue_runners(sess=sess)

[<Thread(Thread-4, started daemon 139861159835392)>,
 <Thread(Thread-5, started daemon 139861147252480)>,
 <Thread(Thread-6, started daemon 139861138859776)>,
 　．．．．．
 <Thread(Thread-35, started daemon 139859318511360)>,
 <Thread(Thread-36, started daemon 139859310118656)>,
 <Thread(Thread-37, started daemon 139859301725952)>]

保存网络结构可视化的二进制文件

#writer = tf.summary.FileWriter('./VggNet_visual',sess.graph)

加载上次训练的参数变量

# 加载训练3200次的权重变量
saver = tf.train.Saver()
saver.restore(sess,'./vgg_weights-3200')

INFO:tensorflow:Restoring parameters from ./vgg_weights-3200

train_cost = training(sess,17８00,5,0.7,10)

Epoch 10: 474 samples/sec, 0.5400 sec/batch, Cost : 0.30948
Epoch 20: 482 samples/sec, 0.5302 sec/batch, Cost : 0.38076
Epoch 30: 479 samples/sec, 0.5339 sec/batch, Cost : 0.33604
　．．．．．．．．．．
Epoch 17500: 473 samples/sec, 0.5411 sec/batch, Cost : 0.23114

fig,ax = plt.subplots(figsize=(13,6))
ax.plot(train_cost)
plt.title('Train Cost')
plt.grid()
plt.show()

前三千次训练

后１７８００次训练

15000次以后已经下降很缓慢了，但是下降的趋势还很明显．

# 训练评估
train__images, train__labels = sess.run([train_images, train_labels])
train_output = sess.run(output,feed_dict={images:train__images,keep_prob:1.0})
train_accuracy = sess.run(accuracy(train__labels, output=train_output))

# 测试评估
test__images, test__labels = sess.run([test_images, test_labels])
test_output = sess.run(output, feed_dict={images:test__images, keep_prob:1.0})
test_accuracy = sess.run(accuracy(test__labels, test_output))

print 'train accuracy is: %.7f'%train_accuracy
print 'test  accuracy is: %.7f'%test_accuracy

train accuracy is: 0.8789062
test  accuracy is: 0.8807812

保存训练３２００次的参数变量

ckpt_dir = './vgg_weights'
saver = tf.train.Saver()
saver.save(sess,save_path=ckpt_dir,global_step=3200)

'./vgg_weights-３２00'

网络结构的可视化

深度卷积网络的论文：
Karen Simonyan,Andrew Zisserman的论文＜＜Very Deep Convolutional Networks for Large-Scale Visual Recognition＞＞http://www.robots.ox.ac.uk/~vgg/research/very_deep/