练习6:ResNet练习

深度学习–ResNet练习

练习6
时间:2020.8.22

昂:

• BasicBlock:

  • conv1 + bn1 + relu
  • conv2 + bn2
  • Identity 注意维度一致
    

• build_resblock:

  • 第一个BasicBlock: 给一个下采样功能
  • 其他的BasicBlock: 不进行下采样功能
    

• ResNet:

  • 一个根部: 第一层
  • resblock堆叠: 根据不同的ResNet给予不同的layer_dims进行堆叠
  • 分类的fc: h, w 未知搞个均值拉好

• 网络:

  • init: 写结构
  • cell: 给数据, 用结构返回输出
  • 重复的分块

• resnet18:
  

• resnet34:

Something:

也许少了点什么,待补0.0

代码:

import  tensorflow as tf
from    tensorflow import keras
from    tensorflow.keras import layers, Sequential

class BasicBlock(layers.Layer):
    
    # filter_num: 卷积核的数量，和channel一个意思
    def __init__(self, filter_num, stride=1):
        super(BasicBlock, self).__init__()
        
        # 第一层:
        self.conv1 = layers.Conv2D(filter_num, (3, 3), strides=stride, padding='same')
        # http://blog.sina.com.cn/s/blog_628cc2b70102yg6w.html
        self.bn1 = layers.BatchNormalization()
        self.relu = layers.Activation('relu')
        
        # 第二层:
        self.conv2 = layers.Conv2D(filter_num, (3, 3), strides=1, padding='same')
        self.bn2 = layers.BatchNormalization()
        
        # Identity 注意保持维度一致
        if stride != 1:
            self.downsample = Sequential()
            self.downsample.add(layers.Conv2D(filter_num, (1, 1), strides=stride)) # w,h,通道数一致
        else:
            self.downsample = lambda x:x

    def call(self, inputs, training=None):

        # [b, h, w, c]
        out = self.conv1(inputs)
        out = self.bn1(out,training=training)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out,training=training)

        identity = self.downsample(inputs)

        output = layers.add([out, identity])
        output = tf.nn.relu(output)

        return output


class ResNet(keras.Model):

    def __init__(self, layer_dims, num_classes=100): # [2, 2, 2, 2]
        super(ResNet, self).__init__()
        
        # 根部:
        self.stem = Sequential([layers.Conv2D(64, (3, 3), strides=(1, 1)),
                                layers.BatchNormalization(),
                                layers.Activation('relu'),
                                layers.MaxPool2D(pool_size=(2, 2), strides=(1, 1), padding='same')
                                ])

        self.layer1 = self.build_resblock(64,  layer_dims[0])
        self.layer2 = self.build_resblock(128, layer_dims[1], stride=2)
        self.layer3 = self.build_resblock(256, layer_dims[2], stride=2)
        self.layer4 = self.build_resblock(512, layer_dims[3], stride=2)
        
        
        # output: [b, 512, h, w] => [b, 512, 1, 1], 无论h,w是多少,在每个channel所有像素值加起来做一个均值
        self.avgpool = layers.GlobalAveragePooling2D()
        # 分类:
        self.fc = layers.Dense(num_classes)

    def call(self, inputs, training=None):

        x = self.stem(inputs,training=training)

        x = self.layer1(x,training=training)
        x = self.layer2(x,training=training)
        x = self.layer3(x,training=training)
        x = self.layer4(x,training=training)

        # [b, c]
        x = self.avgpool(x)
        # [b, 100]
        x = self.fc(x)

        return x


    # blocks: 一个resblock 由几个 BasicBlock 堆叠
    def build_resblock(self, filter_num, blocks, stride=1):
        
        #给第一个block进行下采样功能
        res_blocks = Sequential()
        # may down sample
        res_blocks.add(BasicBlock(filter_num, stride))
        
        #其他的不进行下采样功能
        for _ in range(1, blocks):
            res_blocks.add(BasicBlock(filter_num, stride=1))

        return res_blocks


def resnet18():
    return ResNet([2, 2, 2, 2])


def resnet34():
    return ResNet([3, 4, 6, 3])