注意力机制的原理及实现（pytorch）

本文参加新星计划人工智能(Pytorch)赛道：https://bbs.csdn.net/topics/613989052

空间注意力机制（attention Unet）

class Attention_block(nn.Module):
    def __init__(self, F_g, F_l, F_int):
        super(Attention_block, self).__init__()
        self.W_g = nn.Sequential(
            nn.Conv2d(F_g, F_int, kernel_size=1, stride=1, padding=0, bias=True),
            nn.BatchNorm2d(F_int)
        )

        self.W_x = nn.Sequential(
            nn.Conv2d(F_l, F_int, kernel_size=1, stride=1, padding=0, bias=True),
            nn.BatchNorm2d(F_int)
        )

        self.psi = nn.Sequential(
            nn.Conv2d(F_int, 1, kernel_size=1, stride=1, padding=0, bias=True),
            nn.BatchNorm2d(1),
            nn.Sigmoid()
        )

        self.relu = nn.ReLU(inplace=True)

    def forward(self, g, x):
        # 下采样的gating signal 卷积
        g1 = self.W_g(g)
        # 上采样的 l 卷积
        x1 = self.W_x(x)
        # concat + relu
        psi = self.relu(g1 + x1)
        # channel 减为1，并Sigmoid,得到权重矩阵
        psi = self.psi(psi)
        print(psi.size())
        # 返回加权的 x
        return x * psi

Unet。

通道注意力（seNet）

from torch import nn
class SELayer(nn.Module):
    def __init__(self, channel, reduction=16):
        super(SELayer, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Sequential(
            nn.Linear(channel, channel // reduction, bias=False),
            nn.ReLU(inplace=True),
            nn.Linear(channel // reduction, channel, bias=False),
            nn.Sigmoid()
        )

    def forward(self, x):
        #得到输入张量的batch数量和通道数量
        b, c, _, _ = x.size()
        #通过平均池化，将张量的shape变为1*1
        y = self.avg_pool(x).view(b, c)
        #通过全连接层学习权重，得到通道上的权值
        y = self.fc(y).view(b, c, 1, 1)
        #将y的形状与x对齐并相乘得到最后的输出
        return x * y.expand_as(x)

给定一个输入 ，其特征通道数为 ，通过一系列卷积等一般变换后得到一个特征通道数为 的特征。与传统的CNN不一样的是，接下来通过三个操作来重标定前面得到的特征。

1） Squeeze（压缩）。顺着空间维度来进行特征压缩，将每个二维的特征通道变成一个实数，这个实数某种程度上具有全局的感受野，并且输出的维度和输入的特征通道数相匹配。它表征着在特征通道上响应的全局分布，而且使得靠近输入的层也可以获得全局的感受野，这一点在很多任务中都是非常有用。

2） Excitation（激发）。它是一个类似于循环神经网络中门的机制。通过参数来为每个特征通道生成权重，其中参数被学习用来显式地建模特征通道间的相关性。

3）Reweight（缩放）。将Excitation的输出的权重看做是进过特征选择后的每个特征通道的重要性，然后通过乘法逐通道加权到先前的特征上，完成在通道维度上的对原始特征的重标定

空间注意力+通道注意力（CBAM）

CBAM模块

CAM模块和SAM模块

CBAM的Pytorch实现

# ------------------------#
# CBAM模块的Pytorch实现
# ------------------------#

# 通道注意力模块
class ChannelAttentionModule(nn.Module):
    def __init__(self, channel, reduction=16):  
        super(ChannelAttentionModule, self).__init__()
        mid_channel = channel // reduction
        # 使用自适应池化缩减map的大小，保持通道不变  
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.max_pool = nn.AdaptiveMaxPool2d(1)
        
        self.shared_MLP = nn.Sequential(
            nn.Linear(in_features=channel, out_features=mid_channel),
            nn.ReLU(),
            nn.Linear(in_features=mid_channel, out_features=channel)
        )
        self.sigmoid = nn.Sigmoid()
        # self.act=SiLU()
    
    def forward(self, x):
        avgout = self.shared_MLP(self.avg_pool(x).view(x.size(0),-1)).unsqueeze(2).unsqueeze(3)
        maxout = self.shared_MLP(self.max_pool(x).view(x.size(0),-1)).unsqueeze(2).unsqueeze(3)
        return self.sigmoid(avgout + maxout)
        
# 空间注意力模块
class SpatialAttentionModule(nn.Module):
    def __init__(self):
        super(SpatialAttentionModule, self).__init__()
        self.conv2d = nn.Conv2d(in_channels=2, out_channels=1, kernel_size=7, stride=1, padding=3) 
        # self.act=SiLU()
        self.sigmoid = nn.Sigmoid()
    
    def forward(self, x):
        # map尺寸不变，缩减通道
        avgout = torch.mean(x, dim=1, keepdim=True)
        maxout, _ = torch.max(x, dim=1, keepdim=True)
        out = torch.cat([avgout, maxout], dim=1)
        out = self.sigmoid(self.conv2d(out))
        return out

# CBAM模块
class CBAM(nn.Module):
    def __init__(self, channel): 
        super(CBAM, self).__init__()
        self.channel_attention = ChannelAttentionModule(c1)
        self.spatial_attention = SpatialAttentionModule()

    def forward(self, x):
        out = self.channel_attention(x) * x
        out = self.spatial_attention(out) * out
        return out

ResNet中与一个ResBlock集成的CBAM的用法

pytorch代码实现

# ------------------#
# ResBlock+CBAM
# ------------------#
import torch
import torch.nn as nn
import torchvision


class ChannelAttentionModule(nn.Module):
    def __init__(self, channel, ratio=16):
        super(ChannelAttentionModule, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.max_pool = nn.AdaptiveMaxPool2d(1)

        self.shared_MLP = nn.Sequential(
            nn.Conv2d(channel, channel // ratio, 1, bias=False),
            nn.ReLU(),
            nn.Conv2d(channel // ratio, channel, 1, bias=False)
        )
        self.sigmoid = nn.Sigmoid()

    def forward(self, x):
        avgout = self.shared_MLP(self.avg_pool(x))
        print(avgout.shape)
        maxout = self.shared_MLP(self.max_pool(x))
        return self.sigmoid(avgout + maxout)


class SpatialAttentionModule(nn.Module):
    def __init__(self):
        super(SpatialAttentionModule, self).__init__()
        self.conv2d = nn.Conv2d(in_channels=2, out_channels=1, kernel_size=7, stride=1, padding=3)
        self.sigmoid = nn.Sigmoid()

    def forward(self, x):
        avgout = torch.mean(x, dim=1, keepdim=True)
        maxout, _ = torch.max(x, dim=1, keepdim=True)
        out = torch.cat([avgout, maxout], dim=1)
        out = self.sigmoid(self.conv2d(out))
        return out


class CBAM(nn.Module):
    def __init__(self, channel):
        super(CBAM, self).__init__()
        self.channel_attention = ChannelAttentionModule(channel)
        self.spatial_attention = SpatialAttentionModule()

    def forward(self, x):
        out = self.channel_attention(x) * x
        print('outchannels:{}'.format(out.shape))
        out = self.spatial_attention(out) * out
        return out


class ResBlock_CBAM(nn.Module):
    def __init__(self,in_places, places, stride=1,downsampling=False, expansion = 4):
        super(ResBlock_CBAM,self).__init__()
        self.expansion = expansion
        self.downsampling = downsampling

        self.bottleneck = nn.Sequential(
            nn.Conv2d(in_channels=in_places,out_channels=places,kernel_size=1,stride=1, bias=False),
            nn.BatchNorm2d(places),
            nn.ReLU(inplace=True),
            nn.Conv2d(in_channels=places, out_channels=places, kernel_size=3, stride=stride, padding=1, bias=False),
            nn.BatchNorm2d(places),
            nn.ReLU(inplace=True),
            nn.Conv2d(in_channels=places, out_channels=places*self.expansion, kernel_size=1, stride=1, bias=False),
            nn.BatchNorm2d(places*self.expansion),
        )
        self.cbam = CBAM(channel=places*self.expansion)

        if self.downsampling:
            self.downsample = nn.Sequential(
                nn.Conv2d(in_channels=in_places, out_channels=places*self.expansion, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(places*self.expansion)
            )
        self.relu = nn.ReLU(inplace=True)

    def forward(self, x):
        residual = x
        out = self.bottleneck(x)
        print(x.shape)
        out = self.cbam(out)
        if self.downsampling:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)
        return out


model = ResBlock_CBAM(in_places=16, places=4)
print(model)

input = torch.randn(2, 16, 64, 64)
out = model(input)
print(out.shape)

自注意力机制（self-attention）

# Muti-head Attention 机制的实现
from math import sqrt
import torch
import torch.nn


class Self_Attention(nn.Module):
    # input : batch_size * seq_len * input_dim
    # q : batch_size * input_dim * dim_k
    # k : batch_size * input_dim * dim_k
    # v : batch_size * input_dim * dim_v
    def __init__(self,input_dim,dim_k,dim_v):
        super(Self_Attention,self).__init__()
        self.q = nn.Linear(input_dim,dim_k)
        self.k = nn.Linear(input_dim,dim_k)
        self.v = nn.Linear(input_dim,dim_v)
        self._norm_fact = 1 / sqrt(dim_k)
        
    
    def forward(self,x):
        Q = self.q(x) # Q: batch_size * seq_len * dim_k
        K = self.k(x) # K: batch_size * seq_len * dim_k
        V = self.v(x) # V: batch_size * seq_len * dim_v
         
        atten = nn.Softmax(dim=-1)(torch.bmm(Q,K.permute(0,2,1))) * self._norm_fact # Q * K.T() # batch_size * seq_len * seq_len
        
        output = torch.bmm(atten,V) # Q * K.T() * V # batch_size * seq_len * dim_v
        
        return output

vison transformer实现

input:[2,3,256,256]

划分小patch:patch_szie:32*32,num_patches=(256//32)**2

n,c,w,h --> n,w*h//(32*32), 32*32*c

linear: n,w*h//(32*32), 32*32*c -->n,w*h//(32*32), 32*32

position_embeding

class_token

import torch
from torch import nn

from einops import rearrange, repeat
from einops.layers.torch import Rearrange

# helpers
#返回一个tuple，宽高信息
def pair(t):
    return t if isinstance(t, tuple) else (t, t)

# classes

class PreNorm(nn.Module):
    def __init__(self, dim, fn):
        super().__init__()
        self.norm = nn.LayerNorm(dim)
        self.fn = fn
    def forward(self, x, **kwargs):
        return self.fn(self.norm(x), **kwargs)

class FeedForward(nn.Module):
    def __init__(self, dim, hidden_dim, dropout = 0.):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(dim, hidden_dim),
            nn.ReLU(),
            nn.Dropout(dropout),
            nn.Linear(hidden_dim, dim),
            nn.Dropout(dropout)
        )
    def forward(self, x):
        return self.net(x)

class Attention(nn.Module):
    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
        super().__init__()
        inner_dim = dim_head *  heads
        project_out = not (heads == 1 and dim_head == dim)

        self.heads = heads
        self.scale = dim_head ** -0.5

        self.attend = nn.Softmax(dim = -1)
        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)

        self.to_out = nn.Sequential(
            nn.Linear(inner_dim, dim),
            nn.Dropout(dropout)
        ) if project_out else nn.Identity()

    def forward(self, x):
        qkv = self.to_qkv(x).chunk(3, dim = -1)
        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h = self.heads), qkv)

        dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale

        attn = self.attend(dots)

        out = torch.matmul(attn, v)
        out = rearrange(out, 'b h n d -> b n (h d)')
        return self.to_out(out)

class Transformer(nn.Module):
    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
        super().__init__()
        self.layers = nn.ModuleList([])
        for _ in range(depth):
            self.layers.append(nn.ModuleList([
                PreNorm(dim, Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout)),
                PreNorm(dim, FeedForward(dim, mlp_dim, dropout = dropout))
            ]))
    def forward(self, x):
        for attn, ff in self.layers:
            x = attn(x) + x
            x = ff(x) + x
        return x

class ViT(nn.Module):
    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, pool = 'cls', channels = 3, dim_head = 64, dropout = 0., emb_dropout = 0.):
        super().__init__()
        image_height, image_width = pair(image_size)
        patch_height, patch_width = pair(patch_size)

        assert image_height % patch_height == 0 and image_width % patch_width == 0, 'Image dimensions must be divisible by the patch size.'

        num_patches = (image_height // patch_height) * (image_width // patch_width)
        patch_dim = channels * patch_height * patch_width
        assert pool in {'cls', 'mean'}, 'pool type must be either cls (cls token) or mean (mean pooling)'

        self.to_patch_embedding = nn.Sequential(
            Rearrange('b c (h p1) (w p2) -> b (h w) (p1 p2 c)', p1 = patch_height, p2 = patch_width),
            nn.Linear(patch_dim, dim),
        )

        self.pos_embedding = nn.Parameter(torch.randn(1, num_patches + 1, dim))
        self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
        self.dropout = nn.Dropout(emb_dropout)

        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)

        self.pool = pool
        self.to_latent = nn.Identity()

        self.mlp_head = nn.Sequential(
            nn.LayerNorm(dim),
            nn.Linear(dim, num_classes)
        )

    def forward(self, img):
        x = self.to_patch_embedding(img)
        b, n, _ = x.shape

        cls_tokens = repeat(self.cls_token, '() n d -> b n d', b = b)
        x = torch.cat((cls_tokens, x), dim=1)
        x += self.pos_embedding[:, :(n + 1)]
        x = self.dropout(x)

        x = self.transformer(x)

        x = x.mean(dim = 1) if self.pool == 'mean' else x[:, 0]

        x = self.to_latent(x)
        return self.mlp_head(x)

测试脚本

import torch
from vit_pytorch import ViT
import numpy as np

v = ViT(
    image_size = 256,
    patch_size = 32,
    num_classes = 1000,
    dim = 1024,
    depth = 6,
    heads = 16,
    mlp_dim = 2048,
    dropout = 0.1,
    emb_dropout = 0.1
)

img = torch.randn(2, 3, 256, 256)

preds = v(img) # (1, 1000)

print(preds.shape)
print(np.argmax(preds.detach().numpy(), 1))

空间注意力
https://blog.csdn.net/weixin_37737254/article/details/125863392
代码 https://github.com/Andy-zhujunwen/UNET-ZOO/blob/master/attention_unet.py

通道注意力
https://blog.csdn.net/gaoxueyi551/article/details/120233959
代码 https://github.com/moskomule/senet.pytorch/blob/master/senet/se_module.py

空间注意力加通道注意力（CBAM）：
https://blog.csdn.net/weixin_41790863/article/details/123413303
论文： https://arxiv.org/pdf/1807.06521.pdf

自注意力机制（vision transformer）
论文： https://arxiv.org/pdf/2010.11929.pdf
参考博客：
https://blog.csdn.net/weixin_42392454/article/details/122667271
https://zhuanlan.zhihu.com/p/410776234