''' borrowed from https://github.com/vchoutas/expose/blob/master/expose/models/backbone/hrnet.py ''' import os.path as osp import torch import torch.nn as nn from torchvision.models.resnet import Bottleneck, BasicBlock BN_MOMENTUM = 0.1 def load_HRNet(pretrained=False): hr_net_cfg_dict = { 'use_old_impl': False, 'pretrained_layers': ['*'], 'stage1': { 'num_modules': 1, 'num_branches': 1, 'num_blocks': [4], 'num_channels': [64], 'block': 'BOTTLENECK', 'fuse_method': 'SUM' }, 'stage2': { 'num_modules': 1, 'num_branches': 2, 'num_blocks': [4, 4], 'num_channels': [48, 96], 'block': 'BASIC', 'fuse_method': 'SUM' }, 'stage3': { 'num_modules': 4, 'num_branches': 3, 'num_blocks': [4, 4, 4], 'num_channels': [48, 96, 192], 'block': 'BASIC', 'fuse_method': 'SUM' }, 'stage4': { 'num_modules': 3, 'num_branches': 4, 'num_blocks': [4, 4, 4, 4], 'num_channels': [48, 96, 192, 384], 'block': 'BASIC', 'fuse_method': 'SUM' } } hr_net_cfg = hr_net_cfg_dict model = HighResolutionNet(hr_net_cfg) return model class HighResolutionModule(nn.Module): 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_inchannels self.fuse_method = fuse_method self.num_branches = num_branches self.multi_scale_output = multi_scale_output self.branches = self._make_branches(num_branches, blocks, num_blocks, num_channels) self.fuse_layers = self._make_fuse_layers() self.relu = nn.ReLU(True) def _check_branches(self, num_branches, blocks, num_blocks, num_inchannels, num_channels): if num_branches != len(num_blocks): error_msg = 'NUM_BRANCHES({}) <> NUM_BLOCKS({})'.format( num_branches, len(num_blocks)) raise ValueError(error_msg) if num_branches != len(num_channels): error_msg = 'NUM_BRANCHES({}) <> NUM_CHANNELS({})'.format( num_branches, len(num_channels)) raise ValueError(error_msg) if num_branches != len(num_inchannels): error_msg = 'NUM_BRANCHES({}) <> NUM_INCHANNELS({})'.format( num_branches, len(num_inchannels)) raise ValueError(error_msg) 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), ) 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 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) def _make_branches(self, num_branches, block, num_blocks, num_channels): branches = [] for i in range(num_branches): branches.append( self._make_one_branch(i, block, num_blocks, num_channels)) return nn.ModuleList(branches) def _make_fuse_layers(self): if self.num_branches == 1: return None num_branches = self.num_branches num_inchannels = self.num_inchannels fuse_layers = [] for i in range(num_branches if self.multi_scale_output else 1): fuse_layer = [] for j in range(num_branches): if j > i: 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: fuse_layer.append(None) else: 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) def get_num_inchannels(self): return self.num_inchannels def forward(self, x): if self.num_branches == 1: return [self.branches[0](x[0])] for i in range(self.num_branches): x[i] = self.branches[i](x[i]) x_fuse = [] 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]) x_fuse.append(self.relu(y)) return x_fuse blocks_dict = {'BASIC': BasicBlock, 'BOTTLENECK': Bottleneck} class HighResolutionNet(nn.Module): def __init__(self, cfg, **kwargs): self.inplanes = 64 super(HighResolutionNet, self).__init__() use_old_impl = cfg.get('use_old_impl') self.use_old_impl = use_old_impl # 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.stage1_cfg = cfg.get('stage1', {}) num_channels = self.stage1_cfg['num_channels'][0] block = blocks_dict[self.stage1_cfg['block']] num_blocks = self.stage1_cfg['num_blocks'][0] self.layer1 = self._make_layer(block, num_channels, num_blocks) stage1_out_channel = block.expansion * num_channels self.stage2_cfg = cfg.get('stage2', {}) num_channels = self.stage2_cfg.get('num_channels', (32, 64)) block = blocks_dict[self.stage2_cfg.get('block')] num_channels = [ num_channels[i] * block.expansion for i in range(len(num_channels)) ] stage2_num_channels = num_channels self.transition1 = self._make_transition_layer([stage1_out_channel], num_channels) self.stage2, pre_stage_channels = self._make_stage( self.stage2_cfg, num_channels) self.stage3_cfg = cfg.get('stage3') num_channels = self.stage3_cfg['num_channels'] block = blocks_dict[self.stage3_cfg['block']] num_channels = [ num_channels[i] * block.expansion for i in range(len(num_channels)) ] stage3_num_channels = num_channels self.transition2 = self._make_transition_layer(pre_stage_channels, num_channels) self.stage3, pre_stage_channels = self._make_stage( self.stage3_cfg, num_channels) self.stage4_cfg = cfg.get('stage4') num_channels = self.stage4_cfg['num_channels'] block = blocks_dict[self.stage4_cfg['block']] num_channels = [ num_channels[i] * block.expansion for i in range(len(num_channels)) ] self.transition3 = self._make_transition_layer(pre_stage_channels, num_channels) stage_4_out_channels = num_channels self.stage4, pre_stage_channels = self._make_stage( self.stage4_cfg, num_channels, multi_scale_output=not self.use_old_impl) stage4_num_channels = num_channels self.output_channels_dim = pre_stage_channels self.pretrained_layers = cfg['pretrained_layers'] self.init_weights() self.avg_pooling = nn.AdaptiveAvgPool2d(1) if use_old_impl: in_dims = (2**2 * stage2_num_channels[-1] + 2**1 * stage3_num_channels[-1] + stage_4_out_channels[-1]) else: # TODO: Replace with parameters in_dims = 4 * 384 self.subsample_4 = self._make_subsample_layer( in_channels=stage4_num_channels[0], num_layers=3) self.subsample_3 = self._make_subsample_layer( in_channels=stage2_num_channels[-1], num_layers=2) self.subsample_2 = self._make_subsample_layer( in_channels=stage3_num_channels[-1], num_layers=1) self.conv_layers = self._make_conv_layer(in_channels=in_dims, num_layers=5) def get_output_dim(self): base_output = { f'layer{idx + 1}': val for idx, val in enumerate(self.output_channels_dim) } output = base_output.copy() for key in base_output: output[f'{key}_avg_pooling'] = output[key] output['concat'] = 2048 return output def _make_transition_layer(self, num_channels_pre_layer, num_channels_cur_layer): num_branches_cur = len(num_channels_cur_layer) num_branches_pre = len(num_channels_pre_layer) transition_layers = [] for i in range(num_branches_cur): if i < num_branches_pre: if num_channels_cur_layer[i] != num_channels_pre_layer[i]: 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: transition_layers.append(None) else: conv3x3s = [] for j in range(i + 1 - num_branches_pre): inchannels = num_channels_pre_layer[-1] outchannels = num_channels_cur_layer[i] \ if j == i - num_branches_pre else inchannels conv3x3s.append( nn.Sequential( nn.Conv2d(inchannels, outchannels, 3, 2, 1, bias=False), nn.BatchNorm2d(outchannels), nn.ReLU(inplace=True))) transition_layers.append(nn.Sequential(*conv3x3s)) return nn.ModuleList(transition_layers) 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) def _make_conv_layer(self, in_channels=2048, num_layers=3, num_filters=2048, stride=1): layers = [] for i in range(num_layers): downsample = nn.Conv2d(in_channels, num_filters, stride=1, kernel_size=1, bias=False) layers.append( Bottleneck(in_channels, num_filters // 4, downsample=downsample)) in_channels = num_filters return nn.Sequential(*layers) def _make_subsample_layer(self, in_channels=96, num_layers=3, stride=2): layers = [] for i in range(num_layers): layers.append( nn.Conv2d(in_channels=in_channels, out_channels=2 * in_channels, kernel_size=3, stride=stride, padding=1)) in_channels = 2 * in_channels layers.append(nn.BatchNorm2d(in_channels, momentum=BN_MOMENTUM)) layers.append(nn.ReLU(inplace=True)) return nn.Sequential(*layers) def _make_stage(self, layer_config, num_inchannels, multi_scale_output=True, log=False): 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'] modules = [] for i in range(num_modules): # multi_scale_output is only used last module if not multi_scale_output and i == num_modules - 1: reset_multi_scale_output = False else: reset_multi_scale_output = True modules.append( HighResolutionModule(num_branches, block, num_blocks, num_inchannels, num_channels, fuse_method, reset_multi_scale_output)) modules[-1].log = log num_inchannels = modules[-1].get_num_inchannels() return nn.Sequential(*modules), num_inchannels 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) x_list = [] for i in range(self.stage3_cfg['num_branches']): if self.transition2[i] is not None: if i < self.stage2_cfg['num_branches']: x_list.append(self.transition2[i](y_list[i])) else: x_list.append(self.transition2[i](y_list[-1])) else: x_list.append(y_list[i]) y_list = self.stage3(x_list) x_list = [] for i in range(self.stage4_cfg['num_branches']): if self.transition3[i] is not None: if i < self.stage3_cfg['num_branches']: x_list.append(self.transition3[i](y_list[i])) else: x_list.append(self.transition3[i](y_list[-1])) else: x_list.append(y_list[i]) if not self.use_old_impl: y_list = self.stage4(x_list) output = {} for idx, x in enumerate(y_list): output[f'layer{idx + 1}'] = x feat_list = [] if self.use_old_impl: x3 = self.subsample_3(x_list[1]) x2 = self.subsample_2(x_list[2]) x1 = x_list[3] feat_list = [x3, x2, x1] else: x4 = self.subsample_4(y_list[0]) x3 = self.subsample_3(y_list[1]) x2 = self.subsample_2(y_list[2]) x1 = y_list[3] feat_list = [x4, x3, x2, x1] xf = self.conv_layers(torch.cat(feat_list, dim=1)) xf = xf.mean(dim=(2, 3)) xf = xf.view(xf.size(0), -1) output['concat'] = xf # y_list = self.stage4(x_list) # output['stage4'] = y_list[0] # output['stage4_avg_pooling'] = self.avg_pooling(y_list[0]).view( # *y_list[0].shape[:2]) # concat_outputs = y_list + x_list # output['concat'] = torch.cat([ # self.avg_pooling(tensor).view(*tensor.shape[:2]) # for tensor in concat_outputs], # dim=1) return output def init_weights(self): for m in self.modules(): if isinstance(m, nn.Conv2d): # nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu') nn.init.normal_(m.weight, std=0.001) for name, _ in m.named_parameters(): if name in ['bias']: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.BatchNorm2d): nn.init.constant_(m.weight, 1) nn.init.constant_(m.bias, 0) elif isinstance(m, nn.ConvTranspose2d): nn.init.normal_(m.weight, std=0.001) for name, _ in m.named_parameters(): if name in ['bias']: nn.init.constant_(m.bias, 0) def load_weights(self, pretrained=''): pretrained = osp.expandvars(pretrained) if osp.isfile(pretrained): pretrained_state_dict = torch.load( pretrained, map_location=torch.device("cpu")) need_init_state_dict = {} for name, m in pretrained_state_dict.items(): if (name.split('.')[0] in self.pretrained_layers or self.pretrained_layers[0] == '*'): need_init_state_dict[name] = m missing, unexpected = self.load_state_dict(need_init_state_dict, strict=False) elif pretrained: raise ValueError('{} is not exist!'.format(pretrained))