一、图和代码

上一讲的图中，有stage1

图例如下;

关键是看pose_hrnet.py中PoseHighResolutionNet模块的forward函数

相关部分如下：

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.conv2(x)
        x = self.bn2(x)
        x = self.relu(x)
        x = self.layer1(x)

其中涉及到的定义如下：

    def __init__(self, cfg, **kwargs):
        self.inplanes = 64
        extra = cfg['MODEL']['EXTRA']
        super(PoseHighResolutionNet, self).__init__()

        # stem net
        self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1,
                               bias=False)
        self.bn1 = nn.BatchNorm2d(64, momentum=BN_MOMENTUM)
        self.conv2 = nn.Conv2d(64, 64, kernel_size=3, stride=2, padding=1,
                               bias=False)
        self.bn2 = nn.BatchNorm2d(64, momentum=BN_MOMENTUM)
        self.relu = nn.ReLU(inplace=True)
        self.layer1 = self._make_layer(Bottleneck, 64, 4)

其中，第14行_make_layer定义如下：

    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, momentum=BN_MOMENTUM),
            )

        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)

还有，__init__第14行有个Bottleneck，定义如下：

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, momentum=BN_MOMENTUM)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM)
        self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1,
                               bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion,
                                  momentum=BN_MOMENTUM)
        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

这个Bottleneck，就是个残差网络。关于残差网络，推荐大家好好百度一下，值得明白他的基本想法。

注意两点：

1. Bottleneck和basicblock两回事

2. self.layer1 = self._make_layer(Bottleneck, 64, 4)最后一个参数，说的是，重复4次。

二、魔性的_make_layer

_make_layer定义中的第3行条件，既：

if stride != 1 or self.inplanes != planes * block.expansion:

对于 self.layer1 = self._make_layer(Bottleneck, 64, 4)，满足吗？

首先，stride=1，前一半不满足了。

其次，self.inplanes=64，而planes=64，block.expansion=4.后一半满足了！

那么downsample就会被执行了。

通俗滴说，layer1的输入与输出，特征维度不一样多，为了残差，就需要下采样！

这里的下采样，你看，kernel_size=1，所以，目的不是分辨率，而是特征维数的统一。

看看这些代码，跟示意图，对的上号吗？我觉得稍有差异。不求甚解了，继续往后看！