【CA】《Coordinate Attention for Efficient Mobile Network Design》

在这里插入图片描述

CVPR-2021

github：https://github.com/Andrew-Qibin/CoordAttention

文章目录

1 Background and Motivation
2 Related Work
3 Advantages / Contributions
4 Method
5 Experiments
- 5.1 Datasets and Metrics
- 5.2 Ablation Studies
- 5.3 Comparison with Other Methods
- 5.4 Applications
6 Conclusion

1 Background and Motivation

SENet（【SENet】《Squeeze-and-Excitation Networks》）注意力忽视了 position information

作者 embedding positional information into channel attention，采用 two 1D feature encoding processes，提出 coordinate attention，direction-aware and position-sensitive

在这里插入图片描述

2 Related Work

Mobile Network Architecture
Attention Mechanisms

3 Advantages / Contributions

设计提出轻量级 Coordinate Attention，兼顾 channel 和 spatial
在分类 / 目标检测 / 分割 benchmarks 上验证了其有效性

在这里插入图片描述

4 Method

SE / CBAM / CA 结构对比

在这里插入图片描述
CBAM 虽然有 spatial attention，但是 fail in modeling long-range dependencies that are essential for vision tasks

下面看看 CA 的公式表达

在这里插入图片描述

结合代码看看

import torch
import torch.nn as nn
import math
import torch.nn.functional as Fclass h_sigmoid(nn.Module):def __init__(self, inplace=True):super(h_sigmoid, self).__init__()self.relu = nn.ReLU6(inplace=inplace)def forward(self, x):return self.relu(x + 3) / 6class h_swish(nn.Module):def __init__(self, inplace=True):super(h_swish, self).__init__()self.sigmoid = h_sigmoid(inplace=inplace)def forward(self, x):return x * self.sigmoid(x)class CoordAtt(nn.Module):def __init__(self, inp, oup, reduction=32):super(CoordAtt, self).__init__()self.pool_h = nn.AdaptiveAvgPool2d((None, 1))self.pool_w = nn.AdaptiveAvgPool2d((1, None))mip = max(8, inp // reduction)self.conv1 = nn.Conv2d(inp, mip, kernel_size=1, stride=1, padding=0)self.bn1 = nn.BatchNorm2d(mip)self.act = h_swish()self.conv_h = nn.Conv2d(mip, oup, kernel_size=1, stride=1, padding=0)self.conv_w = nn.Conv2d(mip, oup, kernel_size=1, stride=1, padding=0)def forward(self, x):identity = xn,c,h,w = x.size()x_h = self.pool_h(x) #（n, c, h, 1）x_w = self.pool_w(x).permute(0, 1, 3, 2)  #（n, c, 1, w）-> (n, c, w, 1) 注意这里转化成和 x_h 一样的排列形式了y = torch.cat([x_h, x_w], dim=2)  # 方便拼一起，conv->bn->acty = self.conv1(y)y = self.bn1(y)y = self.act(y) x_h, x_w = torch.split(y, [h, w], dim=2)  # 再拆开x_w = x_w.permute(0, 1, 3, 2)  # 这里转化回去了a_h = self.conv_h(x_h).sigmoid()a_w = self.conv_w(x_w).sigmoid()out = identity * a_w * a_hreturn out