【YOLO改进】主干插入SimAM模块(基于MMYOLO)

SimAM模块

论文链接:https://arxiv.org/abs/2110.06534

将SimAM模块添加到MMYOLO中

1. 将开源代码SimAM.py文件复制到mmyolo/models/plugins目录下

2. 导入MMYOLO用于注册模块的包: from mmyolo.registry import MODELS

3. 确保 class SimAM中的输入维度为in_channels(因为MMYOLO会提前传入输入维度参数，所以要保持参数名的一致)

4. 利用@MODELS.register_module()将“class SimAM(nn.Module)”注册:

5. 修改mmyolo/models/plugins/__init__.py文件

6. 在终端运行:

   python setup.py install

7. 修改对应的配置文件，并且将plugins的参数“type”设置为“SimAM”，可参考【YOLO改进】主干插入注意力机制模块CBAM（基于MMYOLO）-优快云博客

修改后的SimAM.py

import torch
import torch.nn as nn
from mmyolo.registry import MODELS

@MODELS.register_module()
class SimAM(torch.nn.Module):
    def __init__(self, in_channels, e_lambda=1e-4):
        super(SimAM, self).__init__()

        self.activaton = nn.Sigmoid()
        self.e_lambda = e_lambda

    def __repr__(self):
        s = self.__class__.__name__ + '('
        s += ('lambda=%f)' % self.e_lambda)
        return s

    @staticmethod
    def get_module_name():
        return "simam"

    def forward(self, x):
        b, c, h, w = x.size()

        n = w * h - 1

        x_minus_mu_square = (x - x.mean(dim=[2, 3], keepdim=True)).pow(2)
        y = x_minus_mu_square / (4 * (x_minus_mu_square.sum(dim=[2, 3], keepdim=True) / n + self.e_lambda)) + 0.5

        return x * self.activaton(y)


if __name__ == '__main__':
    input = torch.randn(3, 64, 7, 7)
    model = SimAM()
    outputs = model(input)
    print(outputs.shape)

修改后的__init__.py

# Copyright (c) OpenMMLab. All rights reserved.
from .cbam import CBAM
from .Biformer import BiLevelRoutingAttention
from .A2Attention import DoubleAttention
from .CoordAttention import CoordAtt
from .CoTAttention import CoTAttention
from .ECA import ECAAttention
from .EffectiveSE import EffectiveSEModule
from .EMA import EMA
from .GC import GlobalContext
from .GE import GatherExcite
from .MHSA import MHSA
from .ParNetAttention import ParNetAttention
from .PolarizedSelfAttention import ParallelPolarizedSelfAttention
from .S2Attention import S2Attention
from .SE import SEAttention
from .SequentialSelfAttention import SequentialPolarizedSelfAttention
from .SG