Spaces:
Build error
Build error
| # -------------------------------------------------------- | |
| # Pytorch Faster R-CNN and FPN | |
| # Licensed under The MIT License [see LICENSE for details] | |
| # Written by Zheqi He and Xinlei Chen, Yixiao Ge | |
| # https://github.com/yxgeee/pytorch-FPN/blob/master/lib/nets/resnet_v1.py | |
| # -------------------------------------------------------- | |
| from __future__ import absolute_import | |
| from __future__ import division | |
| from __future__ import print_function | |
| import torch | |
| import torch.nn as nn | |
| import torch.nn.functional as F | |
| import math | |
| import torch.utils.model_zoo as model_zoo | |
| __all__ = [ | |
| 'ResNet_FPN', | |
| 'ResNet', | |
| 'resnet18', | |
| 'resnet34', | |
| 'resnet50', | |
| 'resnet101', | |
| 'resnet152'] | |
| model_urls = { | |
| 'resnet18': 'https://s3.amazonaws.com/pytorch/models/resnet18-5c106cde.pth', | |
| 'resnet34': 'https://s3.amazonaws.com/pytorch/models/resnet34-333f7ec4.pth', | |
| 'resnet50': 'https://s3.amazonaws.com/pytorch/models/resnet50-19c8e357.pth', | |
| 'resnet101': 'https://s3.amazonaws.com/pytorch/models/resnet101-5d3b4d8f.pth', | |
| 'resnet152': 'https://s3.amazonaws.com/pytorch/models/resnet152-b121ed2d.pth', | |
| } | |
| def conv3x3(in_planes, out_planes, stride=1): | |
| "3x3 convolution with padding" | |
| return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride, | |
| padding=1, bias=False) | |
| class BasicBlock(nn.Module): | |
| expansion = 1 | |
| 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 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, | |
| stride=stride, | |
| bias=False) # change | |
| self.bn1 = nn.BatchNorm2d(planes) | |
| self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, # change | |
| padding=1, bias=False) | |
| self.bn2 = nn.BatchNorm2d(planes) | |
| self.conv3 = nn.Conv2d(planes, planes * 4, kernel_size=1, bias=False) | |
| self.bn3 = nn.BatchNorm2d(planes * 4) | |
| 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 BuildBlock(nn.Module): | |
| def __init__(self, planes=512): | |
| super(BuildBlock, self).__init__() | |
| self.planes = planes | |
| # Top-down layers, use nn.ConvTranspose2d to replace | |
| # nn.Conv2d+F.upsample? | |
| self.toplayer1 = nn.Conv2d( | |
| 2048, | |
| planes, | |
| kernel_size=1, | |
| stride=1, | |
| padding=0) # Reduce channels | |
| self.toplayer2 = nn.Conv2d( | |
| 512, planes, kernel_size=3, stride=1, padding=1) | |
| self.toplayer3 = nn.Conv2d( | |
| 512, planes, kernel_size=3, stride=1, padding=1) | |
| # Lateral layers | |
| self.latlayer1 = nn.Conv2d( | |
| 1024, planes, kernel_size=1, stride=1, padding=0) | |
| self.latlayer2 = nn.Conv2d( | |
| 512, planes, kernel_size=1, stride=1, padding=0) | |
| def _upsample_add(self, x, y): | |
| _, _, H, W = y.size() | |
| return F.upsample( | |
| x, | |
| size=( | |
| H, | |
| W), | |
| mode='bilinear', | |
| align_corners=True) + y | |
| def forward(self, c3, c4, c5): | |
| # Top-down | |
| p5 = self.toplayer1(c5) | |
| p4 = self._upsample_add(p5, self.latlayer1(c4)) | |
| p4 = self.toplayer2(p4) | |
| p3 = self._upsample_add(p4, self.latlayer2(c3)) | |
| p3 = self.toplayer3(p3) | |
| return p3, p4, p5 | |
| class ResNet(nn.Module): | |
| def __init__(self, block, layers, num_classes=1000): | |
| self.inplanes = 64 | |
| super(ResNet, self).__init__() | |
| # the symbol is referred to fots. | |
| # Conv1 /2 | |
| self.conv1 = nn.Conv2d(1, 64, kernel_size=7, stride=2, padding=3, | |
| bias=False) | |
| self.bn1 = nn.BatchNorm2d(64) | |
| self.relu = nn.ReLU(inplace=True) | |
| # Pool1 /4 | |
| # maxpool different from pytorch-resnet, to match tf-faster-rcnn | |
| self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) | |
| self.layer1 = self._make_layer( | |
| block, 64, layers[0], stride=1) # Res2 /4 | |
| self.layer2 = self._make_layer( | |
| block, 128, layers[1], stride=2) # Res3 /8 | |
| self.layer3 = self._make_layer( | |
| block, 256, layers[2], stride=2) # Res4 /16 | |
| # use stride 1 for the last conv4 layer (same as tf-faster-rcnn) | |
| self.layer4 = self._make_layer( | |
| block, 512, layers[3], stride=2) # Res5 /32 | |
| for m in self.modules(): | |
| if isinstance(m, nn.Conv2d): | |
| n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels | |
| m.weight.data.normal_(0, math.sqrt(2. / n)) | |
| elif isinstance(m, nn.BatchNorm2d): | |
| m.weight.data.fill_(1) | |
| m.bias.data.zero_() | |
| 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), | |
| ) | |
| 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 resnet18(pretrained=False): | |
| """Constructs a ResNet-18 model. | |
| Args: | |
| pretrained (bool): If True, returns a model pre-trained on ImageNet | |
| """ | |
| model = ResNet(BasicBlock, [2, 2, 2, 2]) | |
| if pretrained: | |
| model.load_state_dict(model_zoo.load_url(model_urls['resnet18'])) | |
| return model | |
| def resnet34(pretrained=False): | |
| """Constructs a ResNet-34 model. | |
| Args: | |
| pretrained (bool): If True, returns a model pre-trained on ImageNet | |
| """ | |
| model = ResNet(BasicBlock, [3, 4, 6, 3]) | |
| if pretrained: | |
| model.load_state_dict(model_zoo.load_url(model_urls['resnet34'])) | |
| return model | |
| def resnet50(pretrained=False): | |
| """Constructs a ResNet-50 model. | |
| Args: | |
| pretrained (bool): If True, returns a model pre-trained on ImageNet | |
| """ | |
| model = ResNet(Bottleneck, [3, 4, 6, 3]) | |
| if pretrained: | |
| model.load_state_dict(model_zoo.load_url(model_urls['resnet50'])) | |
| return model | |
| def resnet101(pretrained=False): | |
| """Constructs a ResNet-101 model. | |
| Args: | |
| pretrained (bool): If True, returns a model pre-trained on ImageNet | |
| """ | |
| model = ResNet(Bottleneck, [3, 4, 23, 3]) | |
| if pretrained: | |
| model.load_state_dict(model_zoo.load_url(model_urls['resnet101'])) | |
| return model | |
| def resnet152(pretrained=False): | |
| """Constructs a ResNet-152 model. | |
| Args: | |
| pretrained (bool): If True, returns a model pre-trained on ImageNet | |
| """ | |
| model = ResNet(Bottleneck, [3, 8, 36, 3]) | |
| if pretrained: | |
| model.load_state_dict(model_zoo.load_url(model_urls['resnet152'])) | |
| return model | |
| class ResNet_FPN(nn.Module): | |
| def __init__(self, num_layers=50): | |
| super(ResNet_FPN, self).__init__() | |
| self._num_layers = num_layers | |
| self._layers = {} | |
| self._init_head_tail() | |
| self.out_planes = self.fpn.planes | |
| def forward(self, x): | |
| c2 = self.head1(x) | |
| c3 = self.head2(c2) | |
| c4 = self.head3(c3) | |
| c5 = self.head4(c4) | |
| p3, p4, p5 = self.fpn( c3, c4, c5) | |
| # net_conv = [p2, p3, p4, p5] | |
| # return p2, [x, self.resnet.conv1(x), c2] | |
| return p3 | |
| def _init_head_tail(self): | |
| # choose different blocks for different number of layers | |
| if self._num_layers == 50: | |
| self.resnet = resnet50() | |
| elif self._num_layers == 101: | |
| self.resnet = resnet101() | |
| elif self._num_layers == 152: | |
| self.resnet = resnet152() | |
| else: | |
| # other numbers are not supported | |
| raise NotImplementedError | |
| # Build Building Block for FPN | |
| self.fpn = BuildBlock() | |
| self.head1 = nn.Sequential( | |
| self.resnet.conv1, | |
| self.resnet.bn1, | |
| self.resnet.relu, | |
| self.resnet.maxpool, | |
| self.resnet.layer1) # /4 | |
| self.head2 = nn.Sequential(self.resnet.layer2) # /8 | |
| self.head3 = nn.Sequential(self.resnet.layer3) # /16 | |
| self.head4 = nn.Sequential(self.resnet.layer4) # /32 | |
| if __name__=='__main__': | |
| model = ResNet_FPN() | |
| x = torch.randn((2,1,64,256)) | |
| y = model(x) | |
| print(y.shape) |