Deep High-Resolution Representation Learning for Human Pose Estimation

參考文獻：

https://blog.csdn.net/weixin_38090070/article/details/103443423?utm_medium=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-9.control&dist_request_id=&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-9.control

https://blog.csdn.net/fovever_/article/details/108227898?utm_medium=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-16.control&dist_request_id=1328724.6494.16167188600213217&depth_1-utm_source=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-16.control

https://blog.csdn.net/weixin_38715903/article/details/101629781?utm_medium=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-4.control&dist_request_id=&depth_1-utm_source=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-4.control

建構每一個分支的交換單元卷積是4個Residual unit,且每一個殘差單元塊包含3*3卷積單元。

def _make_one_branch(self, branch_index, block, num_blocks, num_channels,
                         stride=1):
        downsample = None
        if stride != 1 or \
           self.num_inchannels[branch_index] != num_channels[branch_index] * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(
                    self.num_inchannels[branch_index],
                    num_channels[branch_index] * block.expansion,
                    kernel_size=1, stride=stride, bias=False
                ),
                nn.BatchNorm2d(
                    num_channels[branch_index] * block.expansion,
                    momentum=BN_MOMENTUM
                ),
            )
建構每一個分支的殘差單元數目；首先進行次元的統一
        layers = []
        layers.append(
            block(
                self.num_inchannels[branch_index],
                num_channels[branch_index],
                stride,
                downsample
            )
        )
        self.num_inchannels[branch_index] = \
            num_channels[branch_index] * block.expansion
下面的代碼完成的是建構其餘殘差單元塊
        for i in range(1, num_blocks[branch_index]):
            layers.append(
                block(
                    self.num_inchannels[branch_index],
                    num_channels[branch_index]
                )
            )

        return nn.Sequential(*layers)