HRnet-W18

### HRNet-W18 模型架构与实现细节 HRNet (High-Resolution Network) 是一种用于计算机视觉任务的神经网络结构，特别适用于人体姿态估计、语义分割和其他密集预测任务。HRNet 的核心思想是在整个前向传播过程中保持高分辨率表示的同时，逐步引入低分辨率表示并进行多次跨分辨率融合[^1]。 #### 模型架构概述 HRNet-W18 属于 HRNet 家族中的轻量级版本之一。“W18” 表示该模型的基础卷积层宽度为 18 维度。其主要特点如下： - **高分辨率分支**：HRNet 始终维持至少一条高分辨率路径，在训练和推理阶段不会丢失精细特征信息。 - **多尺度融合机制**：通过交换机模块（Exchange Block），不同分辨率的特征图可以相互交互，从而增强全局上下文感知能力[^2]。 具体来说，HRNet 结构由多个 stage 构成，每个 stage 都会增加新的低分辨率分支，并持续保留已有分支的信息流。最终，所有分支的结果会被重新上采样到最高分辨率，并拼接在一起作为输出。 #### 实现细节 以下是基于 PyTorch 的 HRNet-W18 模型的一个简化实现框架： ```python import torch.nn as nn 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 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 def make_stage(block, num_inchannels, num_modules, num_branches, block_num, fuse_method='SUM'): modules = [] for _ in range(num_modules): multi_resolutions_module = MultiResolutionModule( block=block, num_branches=num_branches, blocks=block_num, num_blocks=[4]*num_branches, num_inchannels=num_inchannels, num_channels=[18*(2**i) for i in range(num_branches)], fuse_method=fuse_method ) modules.append(multi_resolutions_module) return nn.Sequential(*modules) class HighResolutionNet(nn.Module): def __init__(self, cfg, **kwargs): super(HighResolutionNet, self).__init__() # Stem network self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1, bias=False) self.bn1 = nn.BatchNorm2d(64) self.conv2 = nn.Conv2d(64, 64, kernel_size=3, stride=2, padding=1, bias=False) self.bn2 = nn.BatchNorm2d(64) self.relu = nn.ReLU(inplace=True) # Stages of the HRNet self.stage1_cfg = cfg['STAGE1'] num_channels = self.stage1_cfg['NUM_CHANNELS'][0] block = Bottleneck num_blocks = self.stage1_cfg['NUM_BLOCKS'][0] self.layer1 = self._make_layer(block, 64, num_channels, num_blocks) self.stage2_cfg = cfg['STAGE2'] num_channels = self.stage2_cfg['NUM_CHANNELS'] block = BasicBlock num_blocks = self.stage2_cfg['NUM_BLOCKS'] self.transition1 = self._make_transition_layer([256], num_channels) self.stage2, pre_stage_channels = self._make_stage(self.stage2_cfg, num_channels) ... def _make_layer(self, block, inplanes, planes, blocks, stride=1): layers = [] downsample = None ... return nn.Sequential(*layers) def _make_transition_layer(self, num_channels_pre_layer, num_channels_cur_layer): transition_layers = [] ... return nn.ModuleList(transition_layers) def _make_stage(self, layer_config, num_inchannels, multi_scale_output=True): num_modules = layer_config['NUM_MODULES'] num_branches = layer_config['NUM_BRANCHES'] num_blocks = layer_config['NUM_BLOCKS'] num_channels = layer_config['NUM_CHANNELS'] block = blocks_dict[layer_config['BLOCK']] fuse_method = layer_config['FUSE_METHOD'] return make_stage(block, num_inchannels, num_modules, num_branches, num_blocks, fuse_method), \ [item * block.expansion for item in num_channels] 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) x_list = [] for 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) ... return y_list[-1] cfg_hrnet_w18 = { 'STAGE1': {'NUM_MODULES': 1, 'NUM_RANCHES': 1, 'BLOCK': 'BOTTLENECK', ...}, 'STAGE2': {...}, 'STAGE3': {...}, 'STAGE4': {...} } model = HighResolutionNet(cfg_hrnet_w18) ``` 此代码片段展示了如何构建一个基础版的 HRNet-W18 模型。完整的配置文件通常可以通过官方仓库获取[^3]。