import torch.nn as nn import torch from torch.nn import functional as F from torchvision import models class ContextualModule(nn.Module): def __init__(self, features, out_features=512, sizes=(1, 2, 3, 6)): super(ContextualModule, self).__init__() self.scales = [] self.scales = nn.ModuleList([self._make_scale(features, size) for size in sizes]) self.bottleneck = nn.Conv2d(features * 2, out_features, kernel_size=1) self.relu = nn.ReLU() self.weight_net = nn.Conv2d(features,features,kernel_size=1) def __make_weight(self,feature,scale_feature): weight_feature = feature - scale_feature return F.sigmoid(self.weight_net(weight_feature)) def _make_scale(self, features, size): prior = nn.AdaptiveAvgPool2d(output_size=(size, size)) conv = nn.Conv2d(features, features, kernel_size=1, bias=False) return nn.Sequential(prior, conv) def forward(self, feats): h, w = feats.size(2), feats.size(3) multi_scales = [F.upsample(input=stage(feats), size=(h, w), mode='bilinear') for stage in self.scales] weights = [self.__make_weight(feats,scale_feature) for scale_feature in multi_scales] overall_features = [(multi_scales[0]*weights[0]+multi_scales[1]*weights[1]+multi_scales[2]*weights[2]+multi_scales[3]*weights[3])/(weights[0]+weights[1]+weights[2]+weights[3])]+ [feats] bottle = self.bottleneck(torch.cat(overall_features, 1)) return self.relu(bottle) class CANNet(nn.Module): def __init__(self, load_weights=False): super(CANNet, self).__init__() self.seen = 0 self.context = ContextualModule(512, 512) self.frontend_feat = [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512] self.backend_feat = [512, 512, 512,256,128,64] self.frontend = make_layers(self.frontend_feat) self.backend = make_layers(self.backend_feat,in_channels = 512,batch_norm=True, dilation = True) self.output_layer = nn.Conv2d(64, 1, kernel_size=1) if not load_weights: mod = models.vgg16(pretrained = True) self._initialize_weights() for i in range(len(self.frontend.state_dict().items())): list(self.frontend.state_dict().items())[i][1].data[:] = list(mod.state_dict().items())[i][1].data[:] def forward(self,x): x = self.frontend(x) x = self.context(x) x = self.backend(x) x = self.output_layer(x) return x def _initialize_weights(self): for m in self.modules(): if isinstance(m, nn.Conv2d): nn.init.normal_(m.weight, std=0.01) if m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.BatchNorm2d): nn.init.constant_(m.weight, 1) nn.init.constant_(m.bias, 0) def make_layers(cfg, in_channels = 3,batch_norm=False,dilation = False): if dilation: d_rate = 2 else: d_rate = 1 layers = [] for v in cfg: if v == 'M': layers += [nn.MaxPool2d(kernel_size=2, stride=2)] else: conv2d = nn.Conv2d(in_channels, v, kernel_size=3, padding=d_rate,dilation = d_rate) if batch_norm: layers += [conv2d, nn.BatchNorm2d(v), nn.ReLU(inplace=True)] else: layers += [conv2d, nn.ReLU(inplace=True)] in_channels = v return nn.Sequential(*layers)