HRNet网络代码解读：Deep High
论文地址：
官方源码：
HRNet网络结构
这里引用“太阳花的小绿豆”绘制的一张基于HRNet-32模型的结构图。便于后续理解。
重要的部分写在代码注释里了，阅读的时候注意。
def get_pose_net(cfg, is_train, **kwargs):model = PoseHighResolutionNet(cfg, **kwargs)if is_train and cfg['MODEL']['INIT_WEIGHTS']:model.init_weights(cfg['MODEL']['PRETRAINED'])return model
使用了t类，让我们进入到这个类看一下。
首先看函数：
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)
所对应的stem net为：
# stem netself.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)
这里经过两个卷积bn激活函数的操作，后接一个模块，特征通道数下采样4倍，通道变为256.
其中由(, 64, 4)构建。
函数
让我们看下的具体操作。
def _make_layer(self, block, planes, blocks, stride=1):downsample = Noneif 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))# 通道数由64变为256self.inplanes = planes * block.expansion# self.inplanes = 64 * 4 = 256for i in range(1, blocks):# 重复堆叠三次，不使用downsample，其实这里的downsample操作也并没有进行下采样。# 输入通道数为256，输出通道数也为256# 最后得到特征图的大小为下采样4倍，输出通道256的featuremaplayers.append(block(self.inplanes, planes))return nn.Sequential(*layers)
类中 = 4, self. = 64 != 64 *4 执行操作。注意这里并没有对模型进行下采样，= 1，只是沿用了的名称，叫成了。
的搭建如下代码：
class Bottleneck(nn.Module):expansion = 4def __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 = downsampleself.stride = stridedef forward(self, x):residual = xout = 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 += residualout = self.relu(out)return out
其中的输入为，输出为4倍的。
配置文件
STAGE2:NUM_MODULES: 1NUM_BRANCHES: 2BLOCK: BASICNUM_BLOCKS:- 4- 4NUM_CHANNELS:- 32- 64FUSE_METHOD: SUM
函数
x_list = []# NUM_BRANCHES = 2for i in range(self.stage2_cfg['NUM_BRANCHES']):if self.transition1[i] is not None:x_list.append(self.transition1[i](x))else:x_list.append(x)y_list = self.stage2(x_list)
分为模块和Stage模块，其中模块是为了进行下采样，并联不同下采样倍率的操作，Stage模块则是进行特征融合。由低下采样倍率和高下采样倍率的特征图融合在一起。
所对应的 __ init __ 方法里的代码如下：
self.stage2_cfg = extra['STAGE2']num_channels = self.stage2_cfg['NUM_CHANNELS']# num_channels [32, 64]block = blocks_dict[self.stage2_cfg['BLOCK']]# basic blocknum_channels = [num_channels[i] * block.expansion for i in range(len(num_channels))]# num_channels [32 * 1, 64 * 1]self.transition1 = self._make_transition_layer([256], num_channels)self.stage2, pre_stage_channels = self._make_stage(self.stage2_cfg, num_channels)
函数
由er函数定义
def _make_transition_layer(self, num_channels_pre_layer, num_channels_cur_layer):# num_channels_pre_layer 之前layer层的channels个数，在stage2之前是256# num_channels_cur_layer 现在layer层的channels个数，在stage2为[32, 64]num_branches_cur = len(num_channels_cur_layer) # 2num_branches_pre = len(num_channels_pre_layer) # 1transition_layers = []# 对应图片上Transition1上的两层3 * 3卷积for i in range(num_branches_cur): # i = 0, 1if i < num_branches_pre:if num_channels_cur_layer[i] != num_channels_pre_layer[i]:# 如果通道数不相等，则通过卷积层改变通道数# 如果通道数相等，则无需卷积操作，可以直接使用，接到下面第一个else语句transition_layers.append(nn.Sequential(nn.Conv2d(num_channels_pre_layer[i],num_channels_cur_layer[i],3, 1, 1, bias=False),nn.BatchNorm2d(num_channels_cur_layer[i]),nn.ReLU(inplace=True)))else:# 对应Transition2, 3中不进行卷积操作的分支transition_layers.append(None)else:# 对应Transition模块上多出的那一个分支，使用stride = 2 再进行下采样conv3x3s = []for j in range(i+1-num_branches_pre):# 利用num_channels_pre_layer之前shape最小的特征层来生成新的分支inchannels = num_channels_pre_layer[-1]outchannels = num_channels_cur_layer[i] \if j == i-num_branches_pre else inchannelsconv3x3s.append(nn.Sequential(nn.Conv2d(inchannels, outchannels, 3, 2, 1, bias=False),# 这里卷积进行下采样，stride = 2nn.BatchNorm2d(outchannels),nn.ReLU(inplace=True)))transition_layers.append(nn.Sequential(*conv3x3s))return nn.ModuleList(transition_layers)
Stage函数
由函数定义：
def _make_stage(self, layer_config, num_inchannels,multi_scale_output=True):num_modules = layer_config['NUM_MODULES'] # 1num_branches = layer_config['NUM_BRANCHES'] # 2num_blocks = layer_config['NUM_BLOCKS'] # [4, 4]num_channels = layer_config['NUM_CHANNELS'] # [32, 64]block = blocks_dict[layer_config['BLOCK']] # BasicBlockfuse_method = layer_config['FUSE_METHOD'] # SUMmodules = []# num_modules 表示一个stage中融合进行几次# 最后一次融合是将其他分支的特征融合到最高分辨率的特征图上,只输出最高分辨率的特征图(multi_scale_output = False)# 前几次融合是将所有分支的特征融合到每个特征图上，输出所有尺寸特征图(multi_scale_output = True)for i in range(num_modules):# multi_scale_output is only used last moduleif not multi_scale_output and i == num_modules - 1:reset_multi_scale_output = Falseelse:reset_multi_scale_output = Truemodules.append(HighResolutionModule(num_branches,block,num_blocks,num_inchannels,num_channels,fuse_method,reset_multi_scale_output))num_inchannels = modules[-1].get_num_inchannels()return nn.Sequential(*modules), num_inchannels
函数
def forward(self, x):if self.num_branches == 1:# 如果只有一个分支，则直接将单个分支特征图作为输入送进self.branchesreturn [self.branches[0](x[0])]# 如果有多个分支，则分别将每个分支特征图作为输入送进self.branches[i]，得到x[i]for i in range(self.num_branches):x[i] = self.branches[i](x[i])x_fuse = []# 把不同分支分别进行上采样和下采样然后融合for i in range(len(self.fuse_layers)):y = x[0] if i == 0 else self.fuse_layers[i][0](x[0])for j in range(1, self.num_branches):if i == j:y = y + x[j]else:y = y + self.fuse_layers[i][j](x[j])# 整体部分最后加Relu激活函数x_fuse.append(self.relu(y))return x_fuse
__ init __ 函数
def __init__(self, num_branches, blocks, num_blocks, num_inchannels,num_channels, fuse_method, multi_scale_output=True):super(HighResolutionModule, self).__init__()self._check_branches(num_branches, blocks, num_blocks, num_inchannels, num_channels)self.num_inchannels = num_inchannelsself.fuse_method = fuse_methodself.num_branches = num_branchesself.multi_scale_output = multi_scale_outputself.branches = self._make_branches(num_branches, blocks, num_blocks, num_channels)self.fuse_layers = self._make_fuse_layers()self.relu = nn.ReLU(True)
函数
def _make_branches(self, num_branches, block, num_blocks, num_channels):branches = []# 反复堆叠_make_one_branch，重复num_branches次数for i in range(num_branches):branches.append(self._make_one_branch(i, block, num_blocks, num_channels))return nn.ModuleList(branches)
函数
def _make_one_branch(self, branch_index, block, num_blocks, num_channels,stride=1):downsample = Noneif 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),)layers = []# 第一个layer接入downsample，但是这里不会进行下采样，堆叠一次basicblocklayers.append(block(self.num_inchannels[branch_index],num_channels[branch_index],stride,downsample))# 通道数[32, 64]self.num_inchannels[branch_index] = \num_channels[branch_index] * block.expansion# num_blocks 为 [4, 4]，所以循环次数为3，重复堆叠basicblockfor i in range(1, num_blocks[branch_index]):layers.append(block(self.num_inchannels[branch_index],num_channels[branch_index]))return nn.Sequential(*layers)
以上这部分为堆叠四次，对应图中中左侧的部分。
函数
def _make_fuse_layers(self):if self.num_branches == 1:return Nonenum_branches = self.num_branches # 2num_inchannels = self.num_inchannels # [32, 64]fuse_layers = []# 把j分支的特征融入到i分支中。for i in range(num_branches if self.multi_scale_output else 1):fuse_layer = []for j in range(num_branches):if j > i:# 如果j分支大于i分支，则说明j下采样倍率更高，需要进行上采样与i分支融合。fuse_layer.append(nn.Sequential(nn.Conv2d(num_inchannels[j],num_inchannels[i],1, 1, 0, bias=False),nn.BatchNorm2d(num_inchannels[i]),nn.Upsample(scale_factor=2**(j-i), mode='nearest')))elif j == i:# j分支等于i分支，不需要进行操作fuse_layer.append(None)else:# j分支大于i分支，需要进行下采样，这里stride = 2# 判断k是否是最后一层，不是最后一层需要加Relu激活函数，最后一层则不需要添加conv3x3s = []for k in range(i-j):if k == i - j - 1:num_outchannels_conv3x3 = num_inchannels[i]conv3x3s.append(nn.Sequential(nn.Conv2d(num_inchannels[j],num_outchannels_conv3x3,3, 2, 1, bias=False),nn.BatchNorm2d(num_outchannels_conv3x3)))else:num_outchannels_conv3x3 = num_inchannels[j]conv3x3s.append(nn.Sequential(nn.Conv2d(num_inchannels[j],num_outchannels_conv3x3,3, 2, 1, bias=False),nn.BatchNorm2d(num_outchannels_conv3x3),nn.ReLU(True)))fuse_layer.append(nn.Sequential(*conv3x3s))fuse_layers.append(nn.ModuleList(fuse_layer))return nn.ModuleList(fuse_layers)
self.final_layer = nn.Conv2d(in_channels=pre_stage_channels[0],out_channels=cfg['MODEL']['NUM_JOINTS'],kernel_size=extra['FINAL_CONV_KERNEL'],stride=1,padding=1 if extra['FINAL_CONV_KERNEL'] == 3 else 0)
的参数为k = 1, s = 1, p = 0, = 17对应17个关键点。
后记
其中关键的部分已经再代码中以注释的形式展现，请认真读注释。
另外只介绍了的部分，，4堆叠策略同上，就不再赘述了。
