• 論文地址：SA-Net: Shuffle Attention for Deep Convolutional Neural Networks
• 開(kāi)源代碼：https://github.com/wofmanaf/SA-Net

當(dāng)前的 CNN 中的 attention 機(jī)制主要包括：channel attention 和 spatial attention，當(dāng)前一些方法（GCNet 、CBAM 等）通常將二者集成，容易產(chǎn)生 converging difficulty 和 heavy computation burden 的問(wèn)題。盡管 ECANet 和 SGE 提出了一些優(yōu)化方案，但沒(méi)有充分利用 channel 和 spatial 之間的關(guān)系。因此，作者提出一個(gè)問(wèn)題 “ Can one fuse different attention modules in a lighter but more efficient way? ”
為解決這個(gè)問(wèn)題，作者提出了 shuffle attention，整體框架如下圖所示?？梢钥闯鍪紫葘⑤斎氲奶卣鞣譃?span id="n5n3t3z" class="katex--inline">$g$組，然后每一組的特征進(jìn)行split，分成兩個(gè)分支，分別計(jì)算 channel attention 和 spatial attention，兩種 attention 都使用全連接 + sigmoid 的方法計(jì)算。接著，兩個(gè)分支的結(jié)果拼接到一起，然后合并，得到和輸入尺寸一致的 feature map。 最后，用一個(gè) shuffle 層進(jìn)行處理。

1. 增加ShuffleAttention.yaml文件

# Parameters
nc: 80  # number of classes
depth_multiple: 0.33  # model depth multiple
width_multiple: 0.50  # layer channel multiple
anchors:
  - [10,13, 16,30, 33,23]  # P3/8
  - [30,61, 62,45, 59,119]  # P4/16
  - [116,90, 156,198, 373,326]  # P5/32

# YOLOv5 v6.0 backbone
backbone:
  # [from, number, module, args]
  [[-1, 1, Conv, [64, 6, 2, 2]],  # 0-P1/2
   [-1, 1, Conv, [128, 3, 2]],  # 1-P2/4
   [-1, 3, C3, [128]],
   [-1, 1, Conv, [256, 3, 2]],  # 3-P3/8
   [-1, 6, C3, [256]],
   [-1, 1, Conv, [512, 3, 2]],  # 5-P4/16
   [-1, 9, C3, [512]],
   [-1, 1, Conv, [1024, 3, 2]],  # 7-P5/32
   [-1, 3, C3, [1024]],
   [-1, 1, SPPF, [1024, 5]],  # 9
  ]

# YOLOAir v6.0 head
head:
  [[-1, 1, Conv, [512, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 6], 1, Concat, [1]],  # cat backbone P4
   [-1, 3, C3, [512, False]],  # 13

   [-1, 1, Conv, [256, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 4], 1, Concat, [1]],  # cat backbone P3
   [-1, 3, C3, [256, False]],  # 17 (P3/8-small)

   [-1, 1, Conv, [256, 3, 2]],
   [[-1, 14], 1, Concat, [1]],  # cat head P4
   [-1, 3, C3, [512, False]],  # 20 (P4/16-medium)

   [-1, 1, Conv, [512, 3, 2]],
   [[-1, 10], 1, Concat, [1]],  # cat head P5
   [-1, 3, C3, [1024, False]],  # 23 (P5/32-large)
   [-1, 1, ShuffleAttention, [1024]], # 修改

   [[17, 20, 24], 1, Detect, [nc, anchors]],  # Detect(P3, P4, P5)
  ]

2. common.py配置

在./models/common.py文件中增加以下模塊代碼

import numpy as np
import torch
from torch import nn
from torch.nn import init
from torch.nn.parameter import Parameter

# https://arxiv.org/pdf/2102.00240.pdf
class ShuffleAttention(nn.Module):

    def __init__(self, channel=512,reduction=16,G=8):
        super().__init__()
        self.G=G
        self.channel=channel
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.gn = nn.GroupNorm(channel // (2 * G), channel // (2 * G))
        self.cweight = Parameter(torch.zeros(1, channel // (2 * G), 1, 1))
        self.cbias = Parameter(torch.ones(1, channel // (2 * G), 1, 1))
        self.sweight = Parameter(torch.zeros(1, channel // (2 * G), 1, 1))
        self.sbias = Parameter(torch.ones(1, channel // (2 * G), 1, 1))
        self.sigmoid=nn.Sigmoid()

    def init_weights(self):
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                init.kaiming_normal_(m.weight, mode='fan_out')
                if m.bias is not None:
                    init.constant_(m.bias, 0)
            elif isinstance(m, nn.BatchNorm2d):
                init.constant_(m.weight, 1)
                init.constant_(m.bias, 0)
            elif isinstance(m, nn.Linear):
                init.normal_(m.weight, std=0.001)
                if m.bias is not None:
                    init.constant_(m.bias, 0)

    @staticmethod
    def channel_shuffle(x, groups):
        b, c, h, w = x.shape
        x = x.reshape(b, groups, -1, h, w)
        x = x.permute(0, 2, 1, 3, 4)

        # flatten
        x = x.reshape(b, -1, h, w)

        return x

    def forward(self, x):
        b, c, h, w = x.size()
        #group into subfeatures
        x=x.view(b*self.G,-1,h,w) #bs*G,c//G,h,w

        #channel_split
        x_0,x_1=x.chunk(2,dim=1) #bs*G,c//(2*G),h,w

        #channel attention
        x_channel=self.avg_pool(x_0) #bs*G,c//(2*G),1,1
        x_channel=self.cweight*x_channel+self.cbias #bs*G,c//(2*G),1,1
        x_channel=x_0*self.sigmoid(x_channel)

        #spatial attention
        x_spatial=self.gn(x_1) #bs*G,c//(2*G),h,w
        x_spatial=self.sweight*x_spatial+self.sbias #bs*G,c//(2*G),h,w
        x_spatial=x_1*self.(x_spatial) #bs*G,c//(2*G),h,w

        # concatenate along channel axis
        out=torch.cat([x_channel,x_spatial],dim=1)  #bs*G,c//G,h,w
        out=out.contiguous().view(b,-1,h,w)

        # channel shuffle
        out = self.channel_shuffle(out, 2)
        return out

3. yolo.py配置

找到models/yolo.py文件中parse_model()函數(shù)的for i, (f, n, m, args) in enumerate(d['backbone'] + d['head'])（258行上下）并其循環(huán)內(nèi)添加如下代碼。文章來(lái)源地址http://www.zghlxwxcb.cn/news/detail-426105.html

elif m is ShuffleAttention:
    c1, c2 = ch[f], args[0]
    if c2 != no:
        c2 = make_divisible(c2 * gw, 8)

4. 訓(xùn)練模型

python train.py --cfg ShuffleAttention.yaml

到了這里，關(guān)于YOLO算法改進(jìn)指南【中階改進(jìn)篇】：3.添加SA-Net注意力機(jī)制的文章就介紹完了。如果您還想了解更多內(nèi)容，請(qǐng)?jiān)谟疑辖撬阉鱐OY模板網(wǎng)以前的文章或繼續(xù)瀏覽下面的相關(guān)文章，希望大家以后多多支持TOY模板網(wǎng)！