yolov26改进 | 添加注意力机制篇 | 最新适用于遥感目标检测的注意力机制CAA二次创新C2PSA机制（全网独家首发+附独家网络结构图）
·
开始讲解之前推荐一下我的专栏，本专栏的内容支持(分类、检测、分割、追踪、关键点检测),专栏目前为限时折扣，欢迎大家订阅本专栏，本专栏每周更新5-7篇最新机制，更有包含我所有改进的文件和交流群提供给大家，本人定期在群内分享发表论文方法和经验。
一、本文介绍
本文给大家带来的最新改进机制是PKINet网络提出的CAA注意力机制， 其首先通过平均池化和一个 1x1 卷积获取局部区域的特征。接着应用两个深度分离的条状卷积，一个是水平方向，另一个是垂直方向。这种条状卷积可以模拟大核卷积，但计算成本低，轻量化。这种配置在捕捉桥梁等拉长的物体结构上特别有效。本文将其用于二次创新PSA和另外一种使用方式，本文内容为个人整理，文章内含有代码 + 添加教程 + 使用方式+独家网络结构图（让你发表论文快人一步）。
欢迎大家订阅我的专栏一起学习YOLO！ 
专栏链接：YOLOv26有效涨点专栏包含：Conv、注意力机制、主干/Backbone、损失函数、优化器、后处理等改进机制
目录
二、原理介绍
官方论文地址： 官方论文地址点击此处即可跳转
官方代码地址： 官方代码地址点击此处即可跳转
CAA（Context Anchor Attention）注意力模块的主要内容很少其实，他是集成在PKINet网络中的：
1. 功能：CAA 集成在 PKI 模块中，旨在通过关注远距离像素间的上下文依赖关系，补充多尺度的局部特征，增强中央区域的特征。
2. 机制：
- 首先通过平均池化和一个 1x1 卷积获取局部区域的特征。
- 接着应用两个深度分离的条状卷积，一个是水平方向，另一个是垂直方向。这种条状卷积可以模拟大核卷积，但计算成本低，轻量化。
- 这种配置在捕捉桥梁等拉长的物体结构上特别有效。
3. 感受野扩展：条状卷积的卷积核大小随着模块深度增加而增大，使得 PKINet 能够更有效地建立远距离像素之间的关系，同时不显著增加计算量。
4. 注意力机制：CAA 模块通过对条状卷积输出应用 Sigmoid 函数生成注意力图，并利用该注意力图对特征图进行缩放，增强相关特征，从而赋予 PKINet 强大的上下文信息处理能力，适应各种物体尺度。
CAA 与 PKI 模块结合，使得 PKINet 能够同时捕捉局部和全局上下文信息，这对遥感目标检测任务尤为关键。
三、核心代码
核心代码的使用方式看章节四！
import torch
import torch.nn as nn
from typing import Optional
__all__ = ['CAA', 'C2PSA_CAA']
class ConvModule(nn.Module):
# 代码重构 CSDN Snu77
def __init__(
self,
in_channels: int, # Number of input channels
out_channels: int, # Number of output channels
kernel_size: int, # Kernel size for convolution
stride: int = 1, # Stride
padding: int = 0, # Padding
groups: int = 1, # Number of groups for grouped convolution
norm_cfg: Optional[dict] = None, # Normalization configuration
act_cfg: Optional[dict] = None): # Activation function configuration
super().__init__()
layers = []
# Convolution layer
layers.append(nn.Conv2d(in_channels, out_channels, kernel_size, stride, padding, groups=groups, bias=(norm_cfg is None)))
# Normalization layer
if norm_cfg:
norm_layer = self._get_norm_layer(out_channels, norm_cfg)
layers.append(norm_layer)
# Activation layer
if act_cfg:
act_layer = self._get_act_layer(act_cfg)
layers.append(act_layer)
# Combine all layers into a sequential layer
self.block = nn.Sequential(*layers)
def forward(self, x):
return self.block(x)
# Helper function to retrieve the normalization layer
def _get_norm_layer(self, num_features, norm_cfg):
if norm_cfg['type'] == 'BN':
return nn.BatchNorm2d(num_features, momentum=norm_cfg.get('momentum', 0.1), eps=norm_cfg.get('eps', 1e-5))
# Add other normalization types here if needed
raise NotImplementedError(f"Normalization layer '{norm_cfg['type']}' is not implemented.")
# Helper function to retrieve the activation layer
def _get_act_layer(self, act_cfg):
if act_cfg['type'] == 'ReLU':
return nn.ReLU(inplace=True)
if act_cfg['type'] == 'SiLU':
return nn.SiLU(inplace=True)
# Add other activation types here if needed
raise NotImplementedError(f"Activation layer '{act_cfg['type']}' is not implemented.")
class CAA(nn.Module):
"""Context Anchor Attention"""
def __init__(
self,
channels: int,
h_kernel_size: int = 11,
v_kernel_size: int = 11,
norm_cfg: Optional[dict] = dict(type='BN', momentum=0.03, eps=0.001),
act_cfg: Optional[dict] = dict(type='SiLU'),
):
super().__init__()
self.avg_pool = nn.AvgPool2d(7, 1, 3)
self.conv1 = ConvModule(channels, channels, 1, 1, 0,
norm_cfg=norm_cfg, act_cfg=act_cfg)
self.h_conv = ConvModule(channels, channels, (1, h_kernel_size), 1,
(0, h_kernel_size // 2), groups=channels,
norm_cfg=None, act_cfg=None)
self.v_conv = ConvModule(channels, channels, (v_kernel_size, 1), 1,
(v_kernel_size // 2, 0), groups=channels,
norm_cfg=None, act_cfg=None)
self.conv2 = ConvModule(channels, channels, 1, 1, 0,
norm_cfg=norm_cfg, act_cfg=act_cfg)
self.act = nn.Sigmoid()
def forward(self, x):
attn_factor = self.act(self.conv2(self.v_conv(self.h_conv(self.conv1(self.avg_pool(x))))))
return attn_factor
def autopad(k, p=None, d=1): # kernel, padding, dilation
"""Pad to 'same' shape outputs."""
if d > 1:
k = d * (k - 1) + 1 if isinstance(k, int) else [d * (x - 1) + 1 for x in k] # actual kernel-size
if p is None:
p = k // 2 if isinstance(k, int) else [x // 2 for x in k] # auto-pad
return p
class Conv(nn.Module):
"""Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation)."""
default_act = nn.SiLU() # default activation
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, d=1, act=True):
"""Initialize Conv layer with given arguments including activation."""
super().__init__()
self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p, d), groups=g, dilation=d, bias=False)
self.bn = nn.BatchNorm2d(c2)
self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()
def forward(self, x):
"""Apply convolution, batch normalization and activation to input tensor."""
return self.act(self.bn(self.conv(x)))
def forward_fuse(self, x):
"""Perform transposed convolution of 2D data."""
return self.act(self.conv(x))
class PSABlock(nn.Module):
"""
PSABlock class implementing a Position-Sensitive Attention block for neural networks.
This class encapsulates the functionality for applying multi-head attention and feed-forward neural network layers
with optional shortcut connections.
Attributes:
attn (Attention): Multi-head attention module.
ffn (nn.Sequential): Feed-forward neural network module.
add (bool): Flag indicating whether to add shortcut connections.
Methods:
forward: Performs a forward pass through the PSABlock, applying attention and feed-forward layers.
Examples:
Create a PSABlock and perform a forward pass
"""
def __init__(self, c, attn_ratio=0.5, num_heads=4, shortcut=True) -> None:
"""Initializes the PSABlock with attention and feed-forward layers for enhanced feature extraction."""
super().__init__()
self.attn = CAA(c)
self.ffn = nn.Sequential(Conv(c, c * 2, 1), Conv(c * 2, c, 1, act=False))
self.add = shortcut
def forward(self, x):
"""Executes a forward pass through PSABlock, applying attention and feed-forward layers to the input tensor."""
x = x + self.attn(x) if self.add else self.attn(x)
x = x + self.ffn(x) if self.add else self.ffn(x)
return x
class C2PSA_CAA(nn.Module):
"""
C2PSA module with attention mechanism for enhanced feature extraction and processing.
This module implements a convolutional block with attention mechanisms to enhance feature extraction and processing
capabilities. It includes a series of PSABlock modules for self-attention and feed-forward operations.
Attributes:
c (int): Number of hidden channels.
cv1 (Conv): 1x1 convolution layer to reduce the number of input channels to 2*c.
cv2 (Conv): 1x1 convolution layer to reduce the number of output channels to c.
m (nn.Sequential): Sequential container of PSABlock modules for attention and feed-forward operations.
Methods:
forward: Performs a forward pass through the C2PSA module, applying attention and feed-forward operations.
Notes:
This module essentially is the same as PSA module, but refactored to allow stacking more PSABlock modules.
Examples:
"""
def __init__(self, c1, c2, n=1, e=0.5):
"""Initializes the C2PSA module with specified input/output channels, number of layers, and expansion ratio."""
super().__init__()
assert c1 == c2
self.c = int(c1 * e)
self.cv1 = Conv(c1, 2 * self.c, 1, 1)
self.cv2 = Conv(2 * self.c, c1, 1)
self.m = nn.Sequential(*(PSABlock(self.c, attn_ratio=0.5, num_heads=self.c // 64) for _ in range(n)))
def forward(self, x):
"""Processes the input tensor 'x' through a series of PSA blocks and returns the transformed tensor."""
a, b = self.cv1(x).split((self.c, self.c), dim=1)
b = self.m(b)
return self.cv2(torch.cat((a, b), 1))
if __name__ == "__main__":
# Generating Sample image
image_size = (1, 36, 224, 224)
image = torch.rand(*image_size)
# Model
model = CAA(36)
out = model(image)
print(out.size())四、添加教程
下面的步骤如果你不会或者不想麻烦操作，可以联系作者获得本专栏添加所有项目文件的源代码，可直接训练.
4.1 修改一
第一还是建立文件，我们找到如下ultralytics/nn文件夹下建立一个目录名字呢就是'Addmodules'文件夹！
4.2 修改二
然后在Addmodules文件夹内建立一个新的py文件，将本文章节三中的“核心代码"复制粘贴进去。
4.3 修改三
第二步我们在该目录下创建一个新的py文件名字为'__init__.py'，然后在其内部导入我们的文件，如下图所示。
4.4 修改四
第三步我门中到如下文件'ultralytics/nn/tasks.py'进行导入和注册我们的模块(此处只需要添加一次即可，如果你用我其它的改进机制这里的步骤只需要添加一次)！
4.5 修改五
在'ultralytics/nn/tasks.py'文件内的parse_model方法函数内（位置大概在1500+行左右），按照图示位置添加即可（此处需要自己有一定的判别能力，如果不会可联系作者获得视频教程）。
4.6 修改六
在'ultralytics/nn/tasks.py'文件内的parse_model方法函数内（位置大概在1550+行左右），按照图示位置添加即可，此处一定要对应好位置和缩进否则很容易报错。
elif m in {此处填写本章代码的名字.}:
c2 = ch[f]
args = [c2, *args]五、正式训练
5.1 yaml文件
5.1.1 yaml文件1
训练信息：YOLO26-C2PSA-CAA summary: 262 layers, 2,491,164 parameters, 2,491,164 gradients, 5.8 GFLOPs
# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
# Ultralytics YOLO26 object detection model with P3/8 - P5/32 outputs
# Model docs: https://docs.ultralytics.com/models/yolo26
# Task docs: https://docs.ultralytics.com/tasks/detect
# Parameters
nc: 80 # number of classes
end2end: True # whether to use end-to-end mode
reg_max: 1 # DFL bins
scales: # model compound scaling constants, i.e. 'model=yolo26n.yaml' will call yolo26.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 260 layers, 2,572,280 parameters, 2,572,280 gradients, 6.1 GFLOPs
s: [0.50, 0.50, 1024] # summary: 260 layers, 10,009,784 parameters, 10,009,784 gradients, 22.8 GFLOPs
m: [0.50, 1.00, 512] # summary: 280 layers, 21,896,248 parameters, 21,896,248 gradients, 75.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 392 layers, 26,299,704 parameters, 26,299,704 gradients, 93.8 GFLOPs
x: [1.00, 1.50, 512] # summary: 392 layers, 58,993,368 parameters, 58,993,368 gradients, 209.5 GFLOPs
# YOLO26n backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
- [-1, 1, Conv, [128, 3, 2]] # 1-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2]] # 3-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
- [-1, 2, C3k2, [512, True]]
- [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32
- [-1, 2, C3k2, [1024, True]]
- [-1, 1, SPPF, [1024, 5, 3, True]] # 9
- [-1, 2, C2PSA_CAA, [1024]] # 10
# YOLO26n head
head:
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 6], 1, Concat, [1]] # cat backbone P4
- [-1, 2, C3k2, [512, True]] # 13
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 4], 1, Concat, [1]] # cat backbone P3
- [-1, 2, C3k2, [256, True]] # 16 (P3/8-small)
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 13], 1, Concat, [1]] # cat head P4
- [-1, 2, C3k2, [512, True]] # 19 (P4/16-medium)
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 10], 1, Concat, [1]] # cat head P5
- [-1, 1, C3k2, [1024, True, 0.5, True]] # 22 (P5/32-large)
- [[16, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5)
5.1.2 yaml文件2
训练信息：YOLO26-Att-CAA summary: 270 layers, 2,516,124 parameters, 2,516,124 gradients, 5.9 GFLOPs
# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
# Ultralytics YOLO26 object detection model with P3/8 - P5/32 outputs
# Model docs: https://docs.ultralytics.com/models/yolo26
# Task docs: https://docs.ultralytics.com/tasks/detect
# Parameters
nc: 80 # number of classes
end2end: True # whether to use end-to-end mode
reg_max: 1 # DFL bins
scales: # model compound scaling constants, i.e. 'model=yolo26n.yaml' will call yolo26.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 260 layers, 2,572,280 parameters, 2,572,280 gradients, 6.1 GFLOPs
s: [0.50, 0.50, 1024] # summary: 260 layers, 10,009,784 parameters, 10,009,784 gradients, 22.8 GFLOPs
m: [0.50, 1.00, 512] # summary: 280 layers, 21,896,248 parameters, 21,896,248 gradients, 75.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 392 layers, 26,299,704 parameters, 26,299,704 gradients, 93.8 GFLOPs
x: [1.00, 1.50, 512] # summary: 392 layers, 58,993,368 parameters, 58,993,368 gradients, 209.5 GFLOPs
# YOLO26n backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
- [-1, 1, Conv, [128, 3, 2]] # 1-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2]] # 3-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
- [-1, 2, C3k2, [512, True]]
- [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32
- [-1, 2, C3k2, [1024, True]]
- [-1, 1, SPPF, [1024, 5, 3, True]] # 9
- [-1, 2, C2PSA, [1024]] # 10
# YOLO26n head
head:
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 6], 1, Concat, [1]] # cat backbone P4
- [-1, 2, C3k2, [512, True]] # 13
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 4], 1, Concat, [1]] # cat backbone P3
- [-1, 2, C3k2, [256, True]] # 16 (P3/8-small)
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 13], 1, Concat, [1]] # cat head P4
- [-1, 2, C3k2, [512, True]] # 19 (P4/16-medium)
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 10], 1, Concat, [1]] # cat head P5
- [-1, 1, C3k2, [1024, True, 0.5, True]] # 22 (P5/32-large)
- [16, 1, CAA, []] # 23
# - [19, 1, CAA, []] # 24
# - [22, 1, CAA, []] # 25
# 此处的使用说法注释: 其中上面的三个注意力机制目前仅使用了23层，如果你想使用24层那么就取消掉代码注释，
# 并将下面检测头中的19改为24,如果想使用第25层注意力机制同理，将下面检测头中的22改为25即可。
# 此处用法比较复杂如过不会联系Snu77博主获取视频教程
- [[23, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5)
5.2 训练代码
大家可以创建一个py文件将我给的代码复制粘贴进去，配置好自己的文件路径即可运行。
import warnings
warnings.filterwarnings('ignore')
from ultralytics import YOLO
if __name__ == '__main__':
model = YOLO('模型配置文件地址,也就是5.1你保存到本地文件的地址')
# 如何切换模型版本, 上面的ymal文件可以改为 yolo26s.yaml就是使用的26s,
# 类似某个改进的yaml文件名称为yolo26-XXX.yaml那么如果想使用其它版本就把上面的名称改为yolo26l-XXX.yaml即可（改的是上面YOLO中间的名字不是配置文件的）！
# model.load('yolo26n.pt') # 是否加载预训练权重,科研不建议大家加载否则很难提升精度
model.train(
data=r"数据集文件地址",
# 如果大家任务是其它的'ultralytics/cfg/default.yaml'找到这里修改task可以改成detect, segment, classify, pose
cache=False,
imgsz=640,
epochs=20,
single_cls=False, # 是否是单类别检测
batch=16,
close_mosaic=0,
workers=0,
device='0',
optimizer='MuSGD', # using SGD/MuSGD
# resume=, # 这里是填写last.pt地址
amp=True, # 如果出现训练损失为Nan可以关闭amp
project='runs/train',
name='exp',
)
5.3 训练过程截图
五、本文总结
到此本文的正式分享内容就结束了，在这里给大家推荐我的YOLOv26改进有效涨点专栏，本专栏目前为新开的平均质量分98分，后期我会根据各种最新的前沿顶会进行论文复现，也会对一些老的改进机制进行补充，如果大家觉得本文帮助到你了，订阅本专栏，关注后续更多的更新~
专栏链接：YOLOv26有效涨点专栏包含：Conv、注意力机制、主干/Backbone、损失函数、优化器、后处理等改进机制