MTCNN（三）基于python代码的网络结构更改

朴灿烈づ我的快乐病毒、 2022-05-10 06:44 155阅读 0赞

背景：MTCNN的训练是在python上实现的，我们需要对其结构进行更改。

目的：读懂MTCNN的python代码。

**目录**

一、代码结构

1.1 tensorflow设置与设备设置

1.2 设置placeholder与out\_tensor

1.3 网络saver

1.4 定义相应的网络fun

1.5 用detect\_face函数给出备选框

二、placeholder与out\_tensor

2.1 tf.placeholder

2.2 PNet，RNet，ONet

2.2.1 pnet = PNet(\{'data': image\_pnet\}, mode='test')

2.2.2 out\_tensor\_pnet = pnet.get\_all\_output()

三、tf.train.saver模型的保存与恢复

四、网络的结构定义

Pnet原始结构

Pnet  最终结构

Rnet原始结构

Rnet最终结构

Onet原始结构

Onet最终采用结构

4.1 conv之中的定义

4.2 wd=self.weight\_decay\_coeff

--------------------

# 一、代码结构 #

## 1.1 tensorflow设置与设备设置 ##

import os  
    os.environ['CUDA_VISIBLE_DEVICES']='1'
        ...
        file_paths = get_model_filenames(args.model_dir)
        with tf.device('/gpu:0'):
            with tf.Graph().as_default():
                config = tf.ConfigProto(allow_soft_placement=True)
                with tf.Session(config=config) as sess:

注意,关于GPU的device是在os.environ\['CUDA\_VISIBLE\_DEVICES'\]='1'后面这个变量来更改的，而不是后面的with tf.device('/gpu:0'):，这个需要后面查找什么意思。

## 1.2 设置placeholder与out\_tensor ##

image_pnet = tf.placeholder(
    	tf.float32, [None, None, None, 3])
    pnet = PNet({'data': image_pnet}, mode='test')
    out_tensor_pnet = pnet.get_all_output()
    
    image_rnet = tf.placeholder(tf.float32, [None, 24, 24, 3])
    rnet = RNet({'data': image_rnet}, mode='test')
    out_tensor_rnet = rnet.get_all_output()
    
    image_onet = tf.placeholder(tf.float32, [None, 48, 48, 3])
    onet = ONet({'data': image_onet}, mode='test')
    out_tensor_onet = onet.get_all_output()

## 1.3 网络saver ##

saver_pnet = tf.train.Saver(
    	[v for v in tf.global_variables()
    	 if v.name[0:5] == "pnet/"])
    saver_rnet = tf.train.Saver(
    	[v for v in tf.global_variables()
    	 if v.name[0:5] == "rnet/"])
    saver_onet = tf.train.Saver(
    	[v for v in tf.global_variables()
    	 if v.name[0:5] == "onet/"])
    
    saver_pnet.restore(sess, file_paths[0])

## 1.4 定义相应的网络fun ##

def pnet_fun(img): return sess.run(
    	out_tensor_pnet, feed_dict={image_pnet: img})
    saver_rnet.restore(sess, file_paths[1])
    
    def rnet_fun(img): return sess.run(
    	out_tensor_rnet, feed_dict={image_rnet: img})
    saver_onet.restore(sess, file_paths[2])
    
    def onet_fun(img): return sess.run(
    	out_tensor_onet, feed_dict={image_onet: img})

## 1.5 用detect\_face函数给出备选框 ##

rectangles, points = detect_face(img, args.minsize,
    		 pnet_fun, rnet_fun, onet_fun,
    		 args.threshold, args.factor)

# 二、placeholder与out\_tensor #

## 2.1 tf.placeholder ##

tf.placeholder(dtype, shape=None, name=None)

placeholder，占位符，在tensorflow中类似于函数参数，运行时必须传入值。

image\_pnet = tf.placeholder(tf.float32, \[None, None, None, 3\])意思就是类型为float32类型，四维的数组，最后一个维度为3。

image\_rnet = tf.placeholder(tf.float32, \[None, 24, 24, 3\])

image\_onet = tf.placeholder(tf.float32, \[None, 48, 48, 3\])

根据向量可以看出分别是12，24与48，但是第一层仅仅是在训练的时候用12\*12来训练，所以维度为None

## 2.2 PNet，RNet，ONet ##

引入在from src.mtcnn import PNet, RNet, ONet

### 2.2.1 pnet = PNet(\{'data': image\_pnet\}, mode='test') ###

定义在src/mtcnn.py之中

#src/mtcnn.py
    class PNet(NetWork):
        def setup(self, task='data', reuse=False):
        ...
        if self.mode == 'train':
            ...
        else
            ...
            self.out_put.append(self.get_output())

pnet是具体化的PNet,其中的'data'对应于image\_pnet,其中的mode对应于'train'的else

### 2.2.2 out\_tensor\_pnet = pnet.get\_all\_output() ###

#src/mtcnn.py
    class NetWork(object):
        ...
        def get_all_output(self):
            return self.out_put
        ...
    
        def get_output(self):
            return self.terminals[-1]

self.output是最终定义完网络结构之后的最终的输出。

针对Pnet，输入为image\_pnet，输出为out\_tensor\_pnet

# 三、tf.train.saver模型的保存与恢复 #

[https://www.cnblogs.com/denny402/p/6940134.html][https_www.cnblogs.com_denny402_p_6940134.html]

[https://blog.csdn.net/index20001/article/details/74322198][https_blog.csdn.net_index20001_article_details_74322198]

# 四、网络的结构定义 #

![70][]

## Pnet原始结构 ##

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;">name</td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv1 prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 10*10*10 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool1</td> 
 <td style="vertical-align:top;width:103.7pt;"> Maxpool 2*2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 5*5*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv2 prelu2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv3 prelu3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 1*1*32 </td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 </tr> 
 </tbody> 
</table>

Pnet改进结构

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 10*10*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 5*5*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 1*1*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 </tr> 
 </tbody> 
</table>

Pnet理想结构

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 6*6*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 1*1*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

## Pnet  最终结构 ##

只有3×3的卷积（为保证输出的得分图与输入的映射，需要same与valid）

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;">name</td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv1 prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Valid</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 10*10*10 </td> 
 <td style="vertical-align:top;width:103.7pt;"> pool1_conv1 pool1_prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 5*5*16 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv2 prelu2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Valid</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv3 prelu3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 1*1*32 </td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 </tr> 
 </tbody> 
</table>

注意！代码更改之后stride也变了，所以需要更改tools之中的generateBoundingBox的stride的尺度，及关于stride的映射。

最终训练结构：

<tf.Variable 'pnet/conv1/weights:0' shape=(3, 3, 3, 10) dtype=float32_ref>
 <tf.Variable 'pnet/conv1/biases:0' shape=(10,) dtype=float32_ref>
 <tf.Variable 'pnet/PReLU1/alpha:0' shape=(10,) dtype=float32_ref>
 <tf.Variable 'pnet/pool1_conv1/weights:0' shape=(3, 3, 10, 16) dtype=float32_ref>
 <tf.Variable 'pnet/pool1_conv1/biases:0' shape=(16,) dtype=float32_ref>
 <tf.Variable 'pnet/pool1_PReLU1/alpha:0' shape=(16,) dtype=float32_ref>
 <tf.Variable 'pnet/conv2/weights:0' shape=(3, 3, 16, 32) dtype=float32_ref>
 <tf.Variable 'pnet/conv2/biases:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'pnet/PReLU2/alpha:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'pnet/conv3/weights:0' shape=(3, 3, 32, 32) dtype=float32_ref>
 <tf.Variable 'pnet/conv3/biases:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'pnet/PReLU3/alpha:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-1/weights:0' shape=(1, 1, 32, 2) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-1/biases:0' shape=(2,) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-2/weights:0' shape=(1, 1, 32, 4) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-2/biases:0' shape=(4,) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-3/weights:0' shape=(1, 1, 32, 10) dtype=float32_ref>
 <tf.Variable 'pnet/conv4-3/biases:0' shape=(10,) dtype=float32_ref>

![70 1][]

## Rnet原始结构 ##

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;">name</td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv1 prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 22*22*28 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool1</td> 
 <td style="vertical-align:top;width:103.7pt;"> maxPool &nbsp;3*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 11*11*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv2 prelu2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 9*9*48 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool2</td> 
 <td style="vertical-align:top;width:103.7pt;"> maxPool&nbsp; 3*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 4*4*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv3 prelu3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2*2*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 </tr> 
 </tbody> 
</table>

Rnet改进结构

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 22*22*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 11*11*28 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 9*9*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 5*5*48 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 <td style="vertical-align:top;width:103.7pt;"> &nbsp; </td> 
 </tr> 
 </tbody> 
</table>

## Rnet最终结构 ##

最终训练结构：

all trainable variables:
 <tf.Variable 'rnet/conv1/weights:0' shape=(3, 3, 3, 28) dtype=float32_ref>
 <tf.Variable 'rnet/conv1/biases:0' shape=(28,) dtype=float32_ref>
 <tf.Variable 'rnet/prelu1/alpha:0' shape=(28,) dtype=float32_ref>
 <tf.Variable 'rnet/pool1_conv1/weights:0' shape=(3, 3, 28, 28) dtype=float32_ref>
 <tf.Variable 'rnet/pool1_conv1/biases:0' shape=(28,) dtype=float32_ref>
 <tf.Variable 'rnet/pool1_prelu1/alpha:0' shape=(28,) dtype=float32_ref>
 <tf.Variable 'rnet/conv2/weights:0' shape=(3, 3, 28, 48) dtype=float32_ref>
 <tf.Variable 'rnet/conv2/biases:0' shape=(48,) dtype=float32_ref>
 <tf.Variable 'rnet/prelu2/alpha:0' shape=(48,) dtype=float32_ref>
 <tf.Variable 'rnet/pool2_conv3/weights:0' shape=(3, 3, 48, 48) dtype=float32_ref>
 <tf.Variable 'rnet/pool2_conv3/biases:0' shape=(48,) dtype=float32_ref>
 <tf.Variable 'rnet/poo2_prelu3/alpha:0' shape=(48,) dtype=float32_ref>
 <tf.Variable 'rnet/conv3/weights:0' shape=(3, 3, 48, 64) dtype=float32_ref>
 <tf.Variable 'rnet/conv3/biases:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'rnet/prelu3/alpha:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'rnet/conv4/weights:0' shape=(576, 128) dtype=float32_ref>
 <tf.Variable 'rnet/conv4/biases:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'rnet/prelu4/alpha:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-1/weights:0' shape=(128, 2) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-1/biases:0' shape=(2,) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-2/weights:0' shape=(128, 4) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-2/biases:0' shape=(4,) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-3/weights:0' shape=(128, 10) dtype=float32_ref>
 <tf.Variable 'rnet/conv5-3/biases:0' shape=(10,) dtype=float32_ref>

![70 2][]

## Onet原始结构 ##

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;">name</td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv1 prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 46*46*32 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool1</td> 
 <td style="vertical-align:top;width:103.7pt;"> maxPool &nbsp;3*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 23*23*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv2 prelu2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 21*21*64 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool2</td> 
 <td style="vertical-align:top;width:103.7pt;"> maxPool &nbsp;3*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 10*10*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv3 prelu3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Valid </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 8*8*64 </td> 
 <td style="vertical-align:top;width:103.7pt;">pool3</td> 
 <td style="vertical-align:top;width:103.7pt;">maxPool&nbsp; 2*2</td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">4*4*64</td> 
 <td style="vertical-align:top;width:103.7pt;"> conv4 prelu4 </td> 
 <td style="vertical-align:top;width:103.7pt;">2*2*128</td> 
 <td style="vertical-align:top;width:103.7pt;">1</td> 
 <td style="vertical-align:top;width:103.7pt;">valid</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

## Onet最终采用结构 ##

因其参数量较小，最终采用 （mAP=58.58%）？

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;">name</td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv1 prelu1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">48*48*32</td> 
 <td style="vertical-align:top;width:103.7pt;"> conv2 prelu2 </td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*32</td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv3 prelu3 </td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*64</td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv4_ prelu4_ </td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*64</td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv5_ prelu5_ </td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 6*6*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> conv6_ prelu6_ </td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

<tf.Variable 'onet/conv1/weights:0' shape=(3, 3, 3, 32) dtype=float32_ref>
 <tf.Variable 'onet/conv1/biases:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/prelu1/alpha:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/conv2/weights:0' shape=(3, 3, 32, 32) dtype=float32_ref>
 <tf.Variable 'onet/conv2/biases:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/prelu2/alpha:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/conv3/weights:0' shape=(3, 3, 32, 64) dtype=float32_ref>
 <tf.Variable 'onet/conv3/biases:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/prelu3/alpha:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/conv4_/weights:0' shape=(3, 3, 64, 64) dtype=float32_ref>
 <tf.Variable 'onet/conv4_/biases:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/prelu4_/alpha:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/conv5_/weights:0' shape=(3, 3, 64, 128) dtype=float32_ref>
 <tf.Variable 'onet/conv5_/biases:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/prelu5_/alpha:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/conv6_/weights:0' shape=(3, 3, 128, 128) dtype=float32_ref>
 <tf.Variable 'onet/conv6_/biases:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/prelu6_/alpha:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/conv5/weights:0' shape=(1152, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv5/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu5/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-1/weights:0' shape=(256, 2) dtype=float32_ref>
 <tf.Variable 'onet/conv6-1/biases:0' shape=(2,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-2/weights:0' shape=(256, 4) dtype=float32_ref>
 <tf.Variable 'onet/conv6-2/biases:0' shape=(4,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-3/weights:0' shape=(256, 10) dtype=float32_ref>
 <tf.Variable 'onet/conv6-3/biases:0' shape=(10,) dtype=float32_ref>

Onet改进结构，只有same的3×3卷积，增加参数量可以增加mAP=59.85%

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 6*6*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

Onet layer 8层卷积结构，只有same的3×3卷积mAP=64.19%

<table> 
 <tbody> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> Feature size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Kernel size </td> 
 <td style="vertical-align:top;width:103.7pt;"> Stride </td> 
 <td style="vertical-align:top;width:103.7pt;"> Padding </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*3 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 48*48*32 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;"> Same </td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*64 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 1 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 24*24*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">12*12*256</td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*256</td> 
 <td style="vertical-align:top;width:103.7pt;">1</td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">12*12*256</td> 
 <td style="vertical-align:top;width:103.7pt;">3*3*256</td> 
 <td style="vertical-align:top;width:103.7pt;">1</td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 12*12*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;"> 6*6*256 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 3*3*128 </td> 
 <td style="vertical-align:top;width:103.7pt;"> 2 </td> 
 <td style="vertical-align:top;width:103.7pt;">Same</td> 
 </tr> 
 <tr> 
 <td style="vertical-align:top;width:103.7pt;">3*3*128</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 <td style="vertical-align:top;width:103.7pt;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

最终训练参数

<tf.Variable 'onet/conv1/weights:0' shape=(3, 3, 3, 32) dtype=float32_ref>
 <tf.Variable 'onet/conv1/biases:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/prelu1/alpha:0' shape=(32,) dtype=float32_ref>
 <tf.Variable 'onet/conv2/weights:0' shape=(3, 3, 32, 64) dtype=float32_ref>
 <tf.Variable 'onet/conv2/biases:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/prelu2/alpha:0' shape=(64,) dtype=float32_ref>
 <tf.Variable 'onet/conv3/weights:0' shape=(3, 3, 64, 128) dtype=float32_ref>
 <tf.Variable 'onet/conv3/biases:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/prelu3/alpha:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/conv4_/weights:0' shape=(3, 3, 128, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv4_/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu_/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv5_/weights:0' shape=(3, 3, 256, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv5_/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu5_/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv6_/weights:0' shape=(3, 3, 256, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv6_/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu6_/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv7_/weights:0' shape=(3, 3, 256, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv7_/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu7_/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv4/weights:0' shape=(2, 2, 256, 128) dtype=float32_ref>
 <tf.Variable 'onet/conv4/biases:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/prelu4/alpha:0' shape=(128,) dtype=float32_ref>
 <tf.Variable 'onet/conv5/weights:0' shape=(1152, 256) dtype=float32_ref>
 <tf.Variable 'onet/conv5/biases:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/prelu5/alpha:0' shape=(256,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-1/weights:0' shape=(256, 2) dtype=float32_ref>
 <tf.Variable 'onet/conv6-1/biases:0' shape=(2,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-2/weights:0' shape=(256, 4) dtype=float32_ref>
 <tf.Variable 'onet/conv6-2/biases:0' shape=(4,) dtype=float32_ref>
 <tf.Variable 'onet/conv6-3/weights:0' shape=(256, 10) dtype=float32_ref>
 <tf.Variable 'onet/conv6-3/biases:0' shape=(10,) dtype=float32_ref>

## 4.1 conv之中的定义 ##

#src/mtcnn.py   in class NetWork(object):  
    def conv(self, inp, k_h, k_w, c_o, s_h, s_w, name,
    		 task=None, relu=True, padding='SAME',
    		 group=1, biased=True, wd=None):
    
    	self.validate_padding(padding)
    	c_i = int(inp.get_shape()[-1])
    	assert c_i % group == 0
    	assert c_o % group == 0
    
    	def convolve(i, k): return tf.nn.conv2d(
    		i, k, [1, s_h, s_w, 1], padding=padding)
    	with tf.variable_scope(name) as scope:
    		kernel = self.make_var(
    			'weights', shape=[
    				k_h, k_w, c_i / group, c_o])
    		if group == 1:
    			output = convolve(inp, kernel)
    		else:
    			input_groups = tf.split(inp, group, 3)
    			kernel_groups = tf.split(kernel, group, 3)
    			output_groups = [convolve(i, k) for i, k in
    							 zip(input_groups, kernel_groups)]
    			output = tf.concat(output_groups, 3)
    		if (wd is not None) and (self.mode == 'train'):
    			self.weight_decay[task].append(
    				tf.multiply(tf.nn.l2_loss(kernel), wd))
    		if biased:
    			biases = self.make_var('biases', [c_o])
    			output = tf.nn.bias_add(output, biases)
    		if relu:
    			output = tf.nn.relu(output, name=scope.name)
    		return output

padding='SAME'就是输入输出一样大，‘VALID’就是不进行padding

几个数字分别为卷积核的大小，卷积核的个数，卷积核时的stride

## 4.2 wd=self.weight\_decay\_coeff ##

网络中有的有此语句，有的没有此语句，该语句全都在后几层。

# src/mtcnn.py  in class class NetWork(object): 
    # in    def conv
    if (wd is not None) and (self.mode == 'train'):
    	self.weight_decay[task].append(
    		tf.multiply(tf.nn.l2_loss(kernel), wd))

是对权重的步长的设置，应该对结果没有太大影响。

[https_www.cnblogs.com_denny402_p_6940134.html]: https://www.cnblogs.com/denny402/p/6940134.html
[https_blog.csdn.net_index20001_article_details_74322198]: https://blog.csdn.net/index20001/article/details/74322198
[70]: /images/20220510/087b29a370bd4ea4a3a786adea3c1dd5.png
[70 1]: /images/20220510/2185c622e9c548aea1bcf5c51231b799.png
[70 2]: /images/20220510/3e7b514c33ac49a6afd8ed9e95e42a0a.png