目标检测论文：FoveaBox: Beyond Anchor-based Object Detector及其PyTorch实现

FoveaBox: Beyond Anchor-based Object Detector
PDF: https://arxiv.org/pdf/1904.03797v1.pdf
PyTorch代码: https://github.com/shanglianlm0525/PyTorch-Networks

创新点:
1 提出了fovea的概念,重新定义正负样本,正样本比GT更小,负样本更远离GT.
2 FPN结构不同层负责不同尺度的目标.
3 Anchor-Free

1 FoveaBox网络结构:

2 方法介绍

2-1 尺度划分（Scale Assignment）

FPN输出5个层P3, P4, P5, P6, P7,每个层分别负责尺度为

的,目标,其中 S l S_{l} Sl分别为为 3 2 2 32^{2} 322, 6 4 2 64^{2} 642, 12 8 2 128^{2} 1282, 25 6 2 256^{2} 2562, 51 2 2 512^{2} 5122. η η η将控制每个层预测的目标尺度.如果 η ≤ 2 η \leq \sqrt{2} η≤2 , 每层目标尺度不重叠,如果 η > 2 η>\sqrt{2} η>2 ,不同层预测的目标将会重叠,即同一个目标将会被多个层预测.

2-2 Fovea定义（Object Fovea）

因为目标真实的边框上的点通常远离目标中心,和背景较为接近,因此FoveaBox先计算出目标中心,然后对目标的真实边框向目标中心收缩一点,作为正样本, 同时向外扩展一点,作为负样本,中间的点忽略掉.

positive area (fovea)定义为:

其中
正负样本的划分为σ1 = 0:3;和 σ2 = 0:4

2-3 预测边框（Box Prediction）

FoveaBox 回归每一个 cell 的坐标映射回原始图像之后和对应的 ground truth 的偏移量.

3 实验结果

PyTorch代码:

import torch
import torch.nn as nn
import torchvisiondef Conv3x3ReLU(in_channels,out_channels):return nn.Sequential(nn.Conv2d(in_channels=in_channels,out_channels=out_channels,kernel_size=3,stride=1,padding=1),nn.ReLU6(inplace=True))def locLayer(in_channels,out_channels):return nn.Sequential(Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=3, stride=1, padding=1),)def confLayer(in_channels,out_channels):return nn.Sequential(Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),Conv3x3ReLU(in_channels=in_channels, out_channels=in_channels),nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=3, stride=1, padding=1),)class FoveaBox(nn.Module):def __init__(self, num_classes=80):super(FoveaBox, self).__init__()self.num_classes = num_classesresnet = torchvision.models.resnet50()layers = list(resnet.children())self.layer1 = nn.Sequential(*layers[:5])self.layer2 = nn.Sequential(*layers[5])self.layer3 = nn.Sequential(*layers[6])self.layer4 = nn.Sequential(*layers[7])self.lateral5 = nn.Conv2d(in_channels=2048, out_channels=256, kernel_size=1)self.lateral4 = nn.Conv2d(in_channels=1024, out_channels=256, kernel_size=1)self.lateral3 = nn.Conv2d(in_channels=512, out_channels=256, kernel_size=1)self.upsample4 = nn.ConvTranspose2d(in_channels=256, out_channels=256, kernel_size=4, stride=2, padding=1)self.upsample3 = nn.ConvTranspose2d(in_channels=256, out_channels=256, kernel_size=4, stride=2, padding=1)self.downsample6 = nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=2, padding=1)self.downsample6_relu = nn.ReLU6(inplace=True)self.downsample5 = nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=2, padding=1)self.loc_layer3 = locLayer(in_channels=256,out_channels=4)self.conf_layer3 = confLayer(in_channels=256,out_channels=self.num_classes)self.loc_layer4 = locLayer(in_channels=256, out_channels=4)self.conf_layer4 = confLayer(in_channels=256, out_channels=self.num_classes)self.loc_layer5 = locLayer(in_channels=256, out_channels=4)self.conf_layer5 = confLayer(in_channels=256, out_channels=self.num_classes)self.loc_layer6 = locLayer(in_channels=256, out_channels=4)self.conf_layer6 = confLayer(in_channels=256, out_channels=self.num_classes)self.loc_layer7 = locLayer(in_channels=256, out_channels=4)self.conf_layer7 = confLayer(in_channels=256, out_channels=self.num_classes)self.init_params()def init_params(self):for m in self.modules():if isinstance(m, nn.Conv2d):nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')elif isinstance(m, nn.BatchNorm2d):nn.init.constant_(m.weight, 1)nn.init.constant_(m.bias, 0)def forward(self, x):x = self.layer1(x)c3 =x = self.layer2(x)c4 =x = self.layer3(x)c5 = x = self.layer4(x)p5 = self.lateral5(c5)p4 = self.upsample4(p5) + self.lateral4(c4)p3 = self.upsample3(p4) + self.lateral3(c3)p6 = self.downsample5(p5)p7 = self.downsample6_relu(self.downsample6(p6))loc3 = self.loc_layer3(p3)conf3 = self.conf_layer3(p3)loc4 = self.loc_layer4(p4)conf4 = self.conf_layer4(p4)loc5 = self.loc_layer5(p5)conf5 = self.conf_layer5(p5)loc6 = self.loc_layer6(p6)conf6 = self.conf_layer6(p6)loc7 = self.loc_layer7(p7)conf7 = self.conf_layer7(p7)locs = torch.cat([loc3.permute(0, 2, 3, 1).contiguous().view(loc3.size(0), -1),loc4.permute(0, 2, 3, 1).contiguous().view(loc4.size(0), -1),loc5.permute(0, 2, 3, 1).contiguous().view(loc5.size(0), -1),loc6.permute(0, 2, 3, 1).contiguous().view(loc6.size(0), -1),loc7.permute(0, 2, 3, 1).contiguous().view(loc7.size(0), -1)],dim=1)confs = torch.cat([conf3.permute(0, 2, 3, 1).contiguous().view(conf3.size(0), -1),conf4.permute(0, 2, 3, 1).contiguous().view(conf4.size(0), -1),conf5.permute(0, 2, 3, 1).contiguous().view(conf5.size(0), -1),conf6.permute(0, 2, 3, 1).contiguous().view(conf6.size(0), -1),conf7.permute(0, 2, 3, 1).contiguous().view(conf7.size(0), -1),], dim=1)out = (locs, confs)return outif __name__ == '__main__':model = FoveaBox()print(model)input = torch.randn(1, 3, 800, 800)out = model(input)print(out[0].shape)print(out[1].shape)