通过和resnet18和resnet50理解PyTorch的ResNet模块

resnet和resnext的框架基本相同的，这里先学习下resnet的构建，感觉高度模块化，很方便。本文算是对 PyTorch源码解读之torchvision.modelsResNet代码的详细理解，另外，强烈推荐这位大神的PyTorch的教程！

模型介绍

resnet的模型可以直接通过torchvision导入，可以通过pretrained设定是否导入预训练的参数。

import torchvision
model = torchvision.models.resnet50(pretrained=False)

如果选择导入，resnet50、resnet101和resnet18等的模型函数十分简洁并且只有ResNet的参数不同，只是需要导入预训练参数时，调用load_state_dict加载model_zoo.load_url下载的参数，这里model_urls是一个维护不同模型参数下载地址的字典。

def resnet18(pretrained=False, **kwargs):
    """Constructs a ResNet-18 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet18']))
    return model
def resnet50(pretrained=False, **kwargs):
    """Constructs a ResNet-50 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet50']))
    return model
model_urls = {
    'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
    'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
    'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
    'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
    'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
}

接下来我们看下重点，也就是ResNet，ResNet的组成是:基础模块Bottleneck/Basicblock，通过_make_layer生成四个的大的layer，然后在forward中排序。
__init__的两个重要参数，block和layers，block有两种(Bottleneck/Basicblock)，不同模型调用的类不同在resnet50、resnet101、resnet152中调用的是Bottleneck类，而在resnet18和resnet34中调用的是BasicBlock类，在后面我们详细理解。layers是包含四个元素的列表，每个元素分别是_make_layer生成四个的大的layer的包含的resdual子结构的个数，在resnet50可以看到列表是 [3, 4, 6, 3]。
_make_layer包含四个参数，第一个参数是block的类型，第二个参数planes是输出的channel数，第三个参数blocks每个blocks中包含多少个residual子结构，也就是上述列表layers所存储的数字，第四个参数为步长。

class ResNet(nn.Module):

    def __init__(self, block, layers, num_classes=1000):
        self.inplanes = 64
        super(ResNet, self).__init__()
        self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3,
                               bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.relu = nn.ReLU(inplace=True)
        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
        self.avgpool = nn.AvgPool2d(7, stride=1)
        self.fc = nn.Linear(512 * block.expansion, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
                m.weight.data.normal_(0, math.sqrt(2. / n)) # 卷积参数变量初始化
            elif isinstance(m, nn.BatchNorm2d):
                m.weight.data.fill_(1) # BN参数初始化
                m.bias.data.zero_()

    def _make_layer(self, block, planes, blocks, stride=1):
        downsample = None
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(self.inplanes, planes * block.expansion,
                          kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(planes * block.expansion),
            )
        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample))
        self.inplanes = planes * block.expansion
        for i in range(1, blocks):
            layers.append(block(self.inplanes, planes))

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)

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

        x = self.avgpool(x)
        x = x.view(x.size(0), -1)
        x = self.fc(x)
        return x

接下来我们看下两种block：Bottleneck/Basicblock，他们最重要的是resdual的结构。所有的模型都继承torch.nn.Module，bottleneck改写了__init__和forward()，forward()中的out += residual就是element-wise add的操作。Bottleneck需要理解的有两处：expansion=4和downsample(下采样)。关于下采样的理论我也不清楚，我们后面直接通过代码来理解吧。

class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(Bottleneck, self).__init__()
        self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.conv3 = nn.Conv2d(planes, planes * 4, kernel_size=1, bias=False)
        self.bn3 = nn.BatchNorm2d(planes * 4)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        out = self.relu(out)

        out = self.conv3(out)
        out = self.bn3(out)

        if self.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out

Basicblock的resdual包含两个卷积层，第一层卷积层的kernel=3。

def conv3x3(in_planes, out_planes, stride=1):
    """3x3 convolution with padding"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
                     padding=1, bias=False)

class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(BasicBlock, self).__init__()
        self.conv1 = conv3x3(inplanes, planes, stride)
        self.bn1 = nn.BatchNorm2d(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = conv3x3(planes, planes)
        self.bn2 = nn.BatchNorm2d(planes)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

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

        if self.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out

resnet18模型流程

resnet调用的Resnet参数是model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
Resnet – init()
self.layer1之前的变量初始化不难理解，self.layer1=self._make_layer(block, 64, layers[0])这里block=Basicblock,layer[0]=2
执行_make_layer
downsample = None——if条件不满足，downsample=None
下面构建blocks层Basicblock：
layers=[]——layers.append(Basicblock(64,64,1,downsample=None))
赋值输入channel self.inplanes = planesblock.expansion = 641 = 64
for循环构建剩下的blocks-1个residual,不传downsample.
self.layer2 执行self._make_layer(block, 128, layers[1], stride=2)
downsample=None
显然if条件满足 downsample=nn.Sequential(nn.Conv2d(64,128, kernel_size=1, stride=2, bias=False), nn.BatchNorm2d(128),
)
layers=[]——layers.append(Basicblock(64,128,2,downsample))
self.inplanes = 128*1=128
for循环构建剩下的blocks-1个residual，不传dowmsample.
可以看出接下来layer3和layer4与layer2相似，最终构成resnet18.

总结

从layer2到layer4，每个layer第一个输入会增加一倍channel，所以resdual会采用下采样，而对于每一层而言，channel都是相同的，basicblock.expansion都为1，所以我们看不出其发挥的作用，我们将在resnet50研究下。如下图，这里没找到resnet18，图中的虚线就是downsample，其产生于channel变化的resdual。

resnet50

model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)，可以看出，resnet50采用Bottleneck模块，并且每个大的layer的blocks数量也不同。
layer1=self._make_layer(Bottleneck, 64, 3)
if条件满足，downsample = nn.Sequential(
nn.Conv2d(self.inplanes=64, 64 * 4,
kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(644),)
layers.append(Bottleneck(64,64,1,dowmsample)),bottleneck内经过三个卷积层Conv2d(64,64) Conv2d(64,64) Conv2d(64,644)保证每个block的输出channel是planesexpansion,通过self.inplanes = planesblock.expansion赋值后面block的输入channel也是planes*expansion。

总结

1.resnet18和resnet50所采用的基础block不同，两种block的卷积层分别为两层和三层。
2.每个大层layer中第二到最后一个block的输入输出是相同的
3.每个layer前会采用下采样，因为这个layer的输入和输出不同，所以resdual需要采用下采样。
4.self.inplanes会在layer第一个block后赋值保证后续的block输入和输出相同。
5.关于resnet和下采样的理论我不太清楚，所以以上就是以来来理解一下操作，欢迎指正。