ResNet结构分析

源码

ResNet封装 在torchvision中封装了Resnet的源码，下面通过对ResNet源码的分析进一步了解ResNet网络结构，方便对ResNet结构进行修改，同时学习网络源码的组织方式，方便日后搭建自己的神经网络

源码解析

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

从源码的入口出发，通过 model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs) 构造网络结构，主要分成两个部分即 Bottleneck 和 [3,4,6,3] 由这两个参数共同决定了ResNet50的网络结构 ，当pretrained 为true时，为model加载imageNet中预训练的参数。
这里涉及到了Bottleneck这个类，[3,4,6,3]对应于上图中ResNet50中 conv2_x中 有三个（1164,3364,11256）卷积层的堆叠 ，同理conv3_中有4个（11128,33128,11512）卷积层的堆叠,resnet将卷积层分为4个大层，[3,4,6,3]代表每一个大层中11,33,1*1 卷积层组合的重复次数 总共1（第一个卷积层）+1（第一个池化层）+（3+4+6+3）*3=50层

这里涉及到一个Bottleneck类，可以把一个Bottleneck当成一个基础的block就是对应上图的（11,33,11）卷积核大小的卷积层的组合

解释一下为什么输入Bottleneck之前是56 * 56 * 64 的，因为ResNet接受的图像大小为224 * 224 经过第一层卷积层
self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False)
floor((224-7+23)/2)+1=112
经过第一层池化之后
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
floor（(112+2*1-3)/2）+1=56

因此在输入到Bottleneck之前得到一个56（height）*56(weight)*64(channel)大小的feature map

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 * self.expansion, kernel_size=1, bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

因为1 * 1卷积核不改变feature map的大小，3 * 3卷积核padding=1 也不改变输入feature map的大小，因此经过一个Bottleneck组成的卷积层组操作后feature map大小不会改变

下面看一下Bottleneck的forward函数

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

这里要留意一下downsample，因为feature map的大小不变 但是在经过Bottleneck 之后channel变成了原来的四倍，因此想要和原始的feature map相加 需要将原始的feature map 也变为原来的四倍 ，downsample 作用是residual+当前feature map时将维度统一

接下来分析ResNet 类的具体构成

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):
                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlineari