网站页面设计稿,竞价被恶意点击怎么办,做静态网站,价格低的英文模块出处
[ICLR 25 Submission] [link] UltraLightUNet: Rethinking U-shaped Network with Multi-kernel Lightweight Convolutions for Medical Image Segmentation 模块名称
Grouped Attention Gate (GAG) 模块作用
轻量特征融合 模块结构 模块特点
特征融合前使用Group…模块出处
[ICLR 25 Submission] [link] UltraLightUNet: Rethinking U-shaped Network with Multi-kernel Lightweight Convolutions for Medical Image Segmentation 模块名称
Grouped Attention Gate (GAG) 模块作用
轻量特征融合 模块结构 模块特点
特征融合前使用Group Conv进行处理比标准卷积更加轻量将融合得到的粗特征视为Spatial Attention Map, 并与Encoder特征相乘从而实现名字中Gate的效果相较于特征融合模块也可以视为一种利用辅助信息(Decoder)特征以增强Encoder特征的增强模块 模块代码
import torch
import torch.nn as nn
import torch.nn.functional as Fclass GAG(nn.Module):def __init__(self, F_g, F_l, F_int, kernel_size1, groups1):super(GAG,self).__init__()if kernel_size 1:groups 1self.W_g nn.Sequential(nn.Conv2d(F_g, F_int, kernel_sizekernel_size,stride1,paddingkernel_size//2,groupsgroups, biasTrue),nn.BatchNorm2d(F_int))self.W_x nn.Sequential(nn.Conv2d(F_l, F_int, kernel_sizekernel_size,stride1,paddingkernel_size//2,groupsgroups, biasTrue),nn.BatchNorm2d(F_int))self.psi nn.Sequential(nn.Conv2d(F_int, 1, kernel_size1,stride1,padding0,biasTrue),nn.BatchNorm2d(1),nn.Sigmoid())self.activation nn.ReLU(inplaceTrue)def forward(self,g,x):g1 self.W_g(g)x1 self.W_x(x)psi self.activation(g1x1)psi self.psi(psi)return x*psiif __name__ __main__:x1 torch.randn([1, 64, 44, 44])x2 torch.randn([1, 64, 44, 44])gag GAG(F_g64, F_l64, F_int64//2, kernel_size3, groups64//2)out gag(x1, x2)print(out.shape) # [1, 64, 44, 44]
