123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245 |
- import torch.nn as nn
- from basicsr.utils.registry import ARCH_REGISTRY
- def conv3x3(inplanes, outplanes, stride=1):
- """A simple wrapper for 3x3 convolution with padding.
- Args:
- inplanes (int): Channel number of inputs.
- outplanes (int): Channel number of outputs.
- stride (int): Stride in convolution. Default: 1.
- """
- return nn.Conv2d(inplanes, outplanes, kernel_size=3, stride=stride, padding=1, bias=False)
- class BasicBlock(nn.Module):
- """Basic residual block used in the ResNetArcFace architecture.
- Args:
- inplanes (int): Channel number of inputs.
- planes (int): Channel number of outputs.
- stride (int): Stride in convolution. Default: 1.
- downsample (nn.Module): The downsample module. Default: None.
- """
- expansion = 1 # output channel expansion ratio
- def __init__(self, inplanes, planes, stride=1, downsample=None):
- super(BasicBlock, self).__init__()
- self.conv1 = conv3x3(inplanes, planes, stride)
- self.bn1 = nn.BatchNorm2d(planes)
- self.relu = nn.ReLU(inplace=True)
- self.conv2 = conv3x3(planes, planes)
- self.bn2 = nn.BatchNorm2d(planes)
- 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)
- if self.downsample is not None:
- residual = self.downsample(x)
- out += residual
- out = self.relu(out)
- return out
- class IRBlock(nn.Module):
- """Improved residual block (IR Block) used in the ResNetArcFace architecture.
- Args:
- inplanes (int): Channel number of inputs.
- planes (int): Channel number of outputs.
- stride (int): Stride in convolution. Default: 1.
- downsample (nn.Module): The downsample module. Default: None.
- use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
- """
- expansion = 1 # output channel expansion ratio
- def __init__(self, inplanes, planes, stride=1, downsample=None, use_se=True):
- super(IRBlock, self).__init__()
- self.bn0 = nn.BatchNorm2d(inplanes)
- self.conv1 = conv3x3(inplanes, inplanes)
- self.bn1 = nn.BatchNorm2d(inplanes)
- self.prelu = nn.PReLU()
- self.conv2 = conv3x3(inplanes, planes, stride)
- self.bn2 = nn.BatchNorm2d(planes)
- self.downsample = downsample
- self.stride = stride
- self.use_se = use_se
- if self.use_se:
- self.se = SEBlock(planes)
- def forward(self, x):
- residual = x
- out = self.bn0(x)
- out = self.conv1(out)
- out = self.bn1(out)
- out = self.prelu(out)
- out = self.conv2(out)
- out = self.bn2(out)
- if self.use_se:
- out = self.se(out)
- if self.downsample is not None:
- residual = self.downsample(x)
- out += residual
- out = self.prelu(out)
- return out
- class Bottleneck(nn.Module):
- """Bottleneck block used in the ResNetArcFace architecture.
- Args:
- inplanes (int): Channel number of inputs.
- planes (int): Channel number of outputs.
- stride (int): Stride in convolution. Default: 1.
- downsample (nn.Module): The downsample module. Default: None.
- """
- expansion = 4 # output channel expansion ratio
- 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
- class SEBlock(nn.Module):
- """The squeeze-and-excitation block (SEBlock) used in the IRBlock.
- Args:
- channel (int): Channel number of inputs.
- reduction (int): Channel reduction ration. Default: 16.
- """
- def __init__(self, channel, reduction=16):
- super(SEBlock, self).__init__()
- self.avg_pool = nn.AdaptiveAvgPool2d(1) # pool to 1x1 without spatial information
- self.fc = nn.Sequential(
- nn.Linear(channel, channel // reduction), nn.PReLU(), nn.Linear(channel // reduction, channel),
- nn.Sigmoid())
- def forward(self, x):
- b, c, _, _ = x.size()
- y = self.avg_pool(x).view(b, c)
- y = self.fc(y).view(b, c, 1, 1)
- return x * y
- @ARCH_REGISTRY.register()
- class ResNetArcFace(nn.Module):
- """ArcFace with ResNet architectures.
- Ref: ArcFace: Additive Angular Margin Loss for Deep Face Recognition.
- Args:
- block (str): Block used in the ArcFace architecture.
- layers (tuple(int)): Block numbers in each layer.
- use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
- """
- def __init__(self, block, layers, use_se=True):
- if block == 'IRBlock':
- block = IRBlock
- self.inplanes = 64
- self.use_se = use_se
- super(ResNetArcFace, self).__init__()
- self.conv1 = nn.Conv2d(1, 64, kernel_size=3, padding=1, bias=False)
- self.bn1 = nn.BatchNorm2d(64)
- self.prelu = nn.PReLU()
- self.maxpool = nn.MaxPool2d(kernel_size=2, stride=2)
- self.layer1 = self._make_layer(block, 64, layers[0])
- self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
- self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
- self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
- self.bn4 = nn.BatchNorm2d(512)
- self.dropout = nn.Dropout()
- self.fc5 = nn.Linear(512 * 8 * 8, 512)
- self.bn5 = nn.BatchNorm1d(512)
- # initialization
- for m in self.modules():
- if isinstance(m, nn.Conv2d):
- nn.init.xavier_normal_(m.weight)
- elif isinstance(m, nn.BatchNorm2d) or isinstance(m, nn.BatchNorm1d):
- nn.init.constant_(m.weight, 1)
- nn.init.constant_(m.bias, 0)
- elif isinstance(m, nn.Linear):
- nn.init.xavier_normal_(m.weight)
- nn.init.constant_(m.bias, 0)
- def _make_layer(self, block, planes, num_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),
- )
- layers = []
- layers.append(block(self.inplanes, planes, stride, downsample, use_se=self.use_se))
- self.inplanes = planes
- for _ in range(1, num_blocks):
- layers.append(block(self.inplanes, planes, use_se=self.use_se))
- return nn.Sequential(*layers)
- def forward(self, x):
- x = self.conv1(x)
- x = self.bn1(x)
- x = self.prelu(x)
- x = self.maxpool(x)
- x = self.layer1(x)
- x = self.layer2(x)
- x = self.layer3(x)
- x = self.layer4(x)
- x = self.bn4(x)
- x = self.dropout(x)
- x = x.view(x.size(0), -1)
- x = self.fc5(x)
- x = self.bn5(x)
- return x
|