|
- import math
- from functools import partial
-
- import torch
- import torch.nn as nn
- import torch.utils.model_zoo as model_zoo
-
- __all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50']
-
- model_urls = {
- 'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
- 'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
- 'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
- 'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
- 'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
- }
-
-
- class BasicBlock(nn.Module):
- expansion = 1
-
- def __init__(self, builder, inplanes, planes, stride=1, downsample=None):
- super(BasicBlock, self).__init__()
- self.conv1 = builder.conv3x3(inplanes, planes, stride)
- self.bn1 = builder.batchnorm(planes)
- self.relu = builder.activation(in_place=True)
- self.conv2 = builder.conv3x3(planes, planes)
- self.bn2 = builder.batchnorm(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 Bottleneck(nn.Module):
- expansion = 4
-
- def __init__(self, builder, inplanes, planes, stride=1, downsample=None):
- super(Bottleneck, self).__init__()
- self.conv1 = builder.conv1x1(inplanes, planes)
- self.bn1 = builder.batchnorm(planes)
- self.conv2 = builder.conv3x3(planes, planes, stride=stride)
- self.bn2 = builder.batchnorm(planes)
- self.conv3 = builder.conv1x1(planes, planes * 4)
- self.bn3 = builder.batchnorm(planes * 4)
- self.relu = builder.activation(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 ResNet(nn.Module):
-
- def __init__(self, block, layers, builder, downsample_dense, num_classes=1000):
- self.inplanes = 64
- super(ResNet, self).__init__()
- if builder is None:
- from utils.builder import get_builder
- builder = get_builder()
-
- if downsample_dense:
- self.dowmsample_conv = partial(nn.Conv2d, kernel_size=1, bias=False)
- else:
- self.dowmsample_conv = builder.conv1x1
-
- self.builder = builder
- self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False)
- self.bn1 = builder.batchnorm(64)
- self.relu = builder.activation(inplace=True)
- self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
- self.layer1 = self._make_layer(builder, block, 64, layers[0])
- self.layer2 = self._make_layer(builder, block, 128, layers[1], stride=2)
- self.layer3 = self._make_layer(builder, block, 256, layers[2], stride=2)
- self.layer4 = self._make_layer(builder, block, 512, layers[3], stride=2)
- self.avgpool = nn.AdaptiveAvgPool2d(1)
- self.fc = nn.Linear(512 * block.expansion, num_classes)
-
- for module in self.modules():
- if isinstance(module, nn.Conv2d):
- n = module.kernel_size[0] * module.kernel_size[1] * module.out_channels
- module.weight.data.normal_(0, math.sqrt(2. / n))
- elif isinstance(module, nn.BatchNorm2d):
- if module.weight is not None:
- module.weight.data.fill_(1)
- module.bias.data.zero_()
-
- def _make_layer(self, builder, block, planes, blocks, stride=1):
- downsample = None
- if stride != 1 or self.inplanes != planes * block.expansion:
- downsample = nn.Sequential(
- self.dowmsample_conv(self.inplanes, planes * block.expansion, stride=stride),
- builder.batchnorm(planes * block.expansion)
- )
-
- layers = []
- layers.append(block(builder, self.inplanes, planes, stride, downsample))
- self.inplanes = planes * block.expansion
- for i in range(1, blocks):
- layers.append(block(builder, self.inplanes, planes))
-
- return nn.Sequential(*layers)
-
- def forward(self, x):
- x = self.conv1(x)
- x = self.bn1(x)
- x = self.relu(x)
- x = self.maxpool(x)
-
- x = self.layer1(x)
- x = self.layer2(x)
- x = self.layer3(x)
- x = self.layer4(x)
-
- x = self.avgpool(x)
- x = x.view(x.size(0), -1)
- x = self.fc(x)
- return x
-
-
- def resnet18(pretrained=False, builder=None, **kwargs):
- """Constructs a ResNet-18 model.
-
- Args:
- pretrained (bool): If True, returns a model pre-trained on ImageNet
- """
- model = ResNet(BasicBlock, [2, 2, 2, 2], builder, downsample_dense=True, **kwargs)
- if pretrained:
- model.load_state_dict(model_zoo.load_url(model_urls['resnet18']), False)
- return model
-
-
- def resnet34(pretrained=False, builder=None, **kwargs):
- """Constructs a ResNet-34 model.
-
- Args:
- pretrained (bool): If True, returns a model pre-trained on ImageNet
- """
- model = ResNet(BasicBlock, [3, 4, 6, 3], builder, downsample_dense=True, **kwargs)
- if pretrained:
- model.load_state_dict(model_zoo.load_url(model_urls['resnet34']), False)
- return model
-
-
- def resnet50(pretrained=False, builder=None, **kwargs):
- """Constructs a ResNet-50 model.
-
- Args:
- pretrained (bool): If True, returns a model pre-trained on ImageNet
- """
- model = ResNet(Bottleneck, [3, 4, 6, 3], builder, downsample_dense=False, **kwargs)
- if pretrained:
- model.load_state_dict(model_zoo.load_url(model_urls['resnet50']), False)
- return model
-
-
- if __name__ == "__main__":
- model = resnet18(num_classes=1000, pretrained=True)
- data = torch.randn(1, 3, 224, 224)
- print(model(data).shape)
|