搭建网站一般多少钱,二次元下午茶wordpress,网络平台贷款逾期不还,沈阳做机床的公司网站已有研究表明#xff1a;将注意力模块嵌入到现有CNN中可以带来显著的性能提升。比如#xff0c;SENet、BAM、CBAM、ECANet、GCNet、FcaNet等注意力机制均带来了可观的性能提升。但是#xff0c;目前仍然存在两个具有挑战性的问题需要解决。一是如何有效地获取和利用不同尺度…已有研究表明将注意力模块嵌入到现有CNN中可以带来显著的性能提升。比如SENet、BAM、CBAM、ECANet、GCNet、FcaNet等注意力机制均带来了可观的性能提升。但是目前仍然存在两个具有挑战性的问题需要解决。一是如何有效地获取和利用不同尺度的特征图的空间信息丰富特征空间。二是通道注意力或者或空间注意力只能有效捕获局部信息而不能建立长期的依赖关系。最新的一些方法虽然能有效解决上述问题但是他们同时会带来巨大的计算负担。基于此本文首先提出了一种新颖的轻量且高效的PSA注意力模块。PSA模块可以处理多尺度的输入特征图的空间信息并且能够有效地建立多尺度通道注意力间的长期依赖关系。然后我们将PSA 模块替换掉ResNet网络Bottleneck中的3x3x卷积其余保持不变最后得到了新的EPSAefficient pyramid split attention block.基于EPSA block我们构建了一个新的骨干网络称作:EPSANet。它既可以提供强有力的多尺度特征表示能力。与此同时EPSANet不仅在图像识别任务中的Top-1 Acc大幅度优于现有技术而且在计算参数量上有更加高效。具体效果如下图所示 论文地址https://arxiv.org/pdf/2105.14447v1.pdf
代码地址https://gitcode.com/mirrors/murufeng/epsanet/blob/master/models/epsanet.py
1.是什么
Pyramid Split Attention (PSA)是一种基于注意力机制的模块它可以用于图像分类、目标检测等任务中。PSA模块通过将不同大小的卷积核的卷积结果进行拼接形成一个金字塔状的特征图然后在这个特征图上应用注意力机制以提取更加丰富的特征信息。与其他注意力模块相比PSA模块具有轻量、简单高效等特点可以与ResNet等主流网络结构结合使用提高模型的性能。
2.为什么
1. SE仅仅考虑了通道注意力忽略了空间注意力。 2. BAM和CBAM考虑了通道注意力和空间注意力但仍存在两个最重要的缺点(1)没有捕获不同尺度的空间信息来丰富特征空间。(2)空间注意力仅仅考虑了局部区域的信息而无法建立远距离的依赖。 3. 后续出现的PyConvRes2Net和HS-ResNet都用于解决CBAM的这两个缺点但计算量太大。
基于以上三点分析提出了Pyramid Split Attention。
3.怎么样
3.1网络结构 PSA模块主要通过四个步骤实现
首先利用SPC模块来对通道进行切分然后针对每个通道特征图上的空间信息进行多尺度特征提取;其次利用SEWeight模块提取不同尺度特征图的通道注意力得到每个不同尺度上的通道注意力向量;第三利用Softmax对多尺度通道注意力向量进行特征重新标定得到新的多尺度通道交互之后的注意力权重。第四对重新校准的权重和相应的特征图按元素进行点乘操作输出得到一个多尺度特征信息注意力加权之后的特征图。该特征图多尺度信息表示能力更丰富。
3.2 SPC module 从前面我们了解到PSA的关键在于多尺度特征提取即SPC模块。假设输入为X我们先将其拆分为S部分,然后对不同部分提取不同尺度特征最后将所提取的多尺度特征通过Concat进行拼接。上述过程可以简单描述如下 在上述特征基础上我们对不同部分特征提取注意力权值公式如下 为更好的实现注意力信息交互并融合跨维度信息我们将上述所得注意力向量进行拼接即.然后我们再对所得注意力权值进行归一化定义如下 最后我们即可得到校正后的特征Y F ⊙ att 3.3 框图 3.4 代码实现
SEWeightModule import torch.nn as nnclass SEWeightModule(nn.Module):def __init__(self, channels, reduction16):super(SEWeightModule, self).__init__()self.avg_pool nn.AdaptiveAvgPool2d(1)self.fc1 nn.Conv2d(channels, channels//reduction, kernel_size1, padding0)self.relu nn.ReLU(inplaceTrue)self.fc2 nn.Conv2d(channels//reduction, channels, kernel_size1, padding0)self.sigmoid nn.Sigmoid()def forward(self, x):out self.avg_pool(x)out self.fc1(out)out self.relu(out)out self.fc2(out)weight self.sigmoid(out)return weight
PSAModule
def conv(in_planes, out_planes, kernel_size3, stride1, padding1, dilation1, groups1):standard convolution with paddingreturn nn.Conv2d(in_planes, out_planes, kernel_sizekernel_size, stridestride,paddingpadding, dilationdilation, groupsgroups, biasFalse)def conv1x1(in_planes, out_planes, stride1):1x1 convolutionreturn nn.Conv2d(in_planes, out_planes, kernel_size1, stridestride, biasFalse)class PSAModule(nn.Module):def __init__(self, inplans, planes, conv_kernels[3, 5, 7, 9], stride1, conv_groups[1, 4, 8, 16]):super(PSAModule, self).__init__()self.conv_1 conv(inplans, planes//4, kernel_sizeconv_kernels[0], paddingconv_kernels[0]//2,stridestride, groupsconv_groups[0])self.conv_2 conv(inplans, planes//4, kernel_sizeconv_kernels[1], paddingconv_kernels[1]//2,stridestride, groupsconv_groups[1])self.conv_3 conv(inplans, planes//4, kernel_sizeconv_kernels[2], paddingconv_kernels[2]//2,stridestride, groupsconv_groups[2])self.conv_4 conv(inplans, planes//4, kernel_sizeconv_kernels[3], paddingconv_kernels[3]//2,stridestride, groupsconv_groups[3])self.se SEWeightModule(planes // 4)self.split_channel planes // 4self.softmax nn.Softmax(dim1)def forward(self, x):batch_size x.shape[0]x1 self.conv_1(x)x2 self.conv_2(x)x3 self.conv_3(x)x4 self.conv_4(x)feats torch.cat((x1, x2, x3, x4), dim1)feats feats.view(batch_size, 4, self.split_channel, feats.shape[2], feats.shape[3])x1_se self.se(x1)x2_se self.se(x2)x3_se self.se(x3)x4_se self.se(x4)x_se torch.cat((x1_se, x2_se, x3_se, x4_se), dim1)attention_vectors x_se.view(batch_size, 4, self.split_channel, 1, 1)attention_vectors self.softmax(attention_vectors)feats_weight feats * attention_vectorsfor i in range(4):x_se_weight_fp feats_weight[:, i, :, :]if i 0:out x_se_weight_fpelse:out torch.cat((x_se_weight_fp, out), 1)return out EPSANET
import torch
import torch.nn as nn
import math
from .SE_weight_module import SEWeightModuledef conv(in_planes, out_planes, kernel_size3, stride1, padding1, dilation1, groups1):standard convolution with paddingreturn nn.Conv2d(in_planes, out_planes, kernel_sizekernel_size, stridestride,paddingpadding, dilationdilation, groupsgroups, biasFalse)def conv1x1(in_planes, out_planes, stride1):1x1 convolutionreturn nn.Conv2d(in_planes, out_planes, kernel_size1, stridestride, biasFalse)class PSAModule(nn.Module):def __init__(self, inplans, planes, conv_kernels[3, 5, 7, 9], stride1, conv_groups[1, 4, 8, 16]):super(PSAModule, self).__init__()self.conv_1 conv(inplans, planes//4, kernel_sizeconv_kernels[0], paddingconv_kernels[0]//2,stridestride, groupsconv_groups[0])self.conv_2 conv(inplans, planes//4, kernel_sizeconv_kernels[1], paddingconv_kernels[1]//2,stridestride, groupsconv_groups[1])self.conv_3 conv(inplans, planes//4, kernel_sizeconv_kernels[2], paddingconv_kernels[2]//2,stridestride, groupsconv_groups[2])self.conv_4 conv(inplans, planes//4, kernel_sizeconv_kernels[3], paddingconv_kernels[3]//2,stridestride, groupsconv_groups[3])self.se SEWeightModule(planes // 4)self.split_channel planes // 4self.softmax nn.Softmax(dim1)def forward(self, x):batch_size x.shape[0]x1 self.conv_1(x)x2 self.conv_2(x)x3 self.conv_3(x)x4 self.conv_4(x)feats torch.cat((x1, x2, x3, x4), dim1)feats feats.view(batch_size, 4, self.split_channel, feats.shape[2], feats.shape[3])x1_se self.se(x1)x2_se self.se(x2)x3_se self.se(x3)x4_se self.se(x4)x_se torch.cat((x1_se, x2_se, x3_se, x4_se), dim1)attention_vectors x_se.view(batch_size, 4, self.split_channel, 1, 1)attention_vectors self.softmax(attention_vectors)feats_weight feats * attention_vectorsfor i in range(4):x_se_weight_fp feats_weight[:, i, :, :]if i 0:out x_se_weight_fpelse:out torch.cat((x_se_weight_fp, out), 1)return outclass EPSABlock(nn.Module):expansion 4def __init__(self, inplanes, planes, stride1, downsampleNone, norm_layerNone, conv_kernels[3, 5, 7, 9],conv_groups[1, 4, 8, 16]):super(EPSABlock, self).__init__()if norm_layer is None:norm_layer nn.BatchNorm2d# Both self.conv2 and self.downsample layers downsample the input when stride ! 1self.conv1 conv1x1(inplanes, planes)self.bn1 norm_layer(planes)self.conv2 PSAModule(planes, planes, stridestride, conv_kernelsconv_kernels, conv_groupsconv_groups)self.bn2 norm_layer(planes)self.conv3 conv1x1(planes, planes * self.expansion)self.bn3 norm_layer(planes * self.expansion)self.relu nn.ReLU(inplaceTrue)self.downsample downsampleself.stride stridedef forward(self, x):identity xout self.conv1(x)out self.bn1(out)out self.relu(out)out self.conv2(out)out self.bn2(out)out self.relu(out)out self.conv3(out)out self.bn3(out)if self.downsample is not None:identity self.downsample(x)out identityout self.relu(out)return outclass EPSANet(nn.Module):def __init__(self,block, layers, num_classes1000):super(EPSANet, self).__init__()self.inplanes 64self.conv1 nn.Conv2d(3, 64, kernel_size7, stride2, padding3, biasFalse)self.bn1 nn.BatchNorm2d(64)self.relu nn.ReLU(inplaceTrue)self.maxpool nn.MaxPool2d(kernel_size3, stride2, padding1)self.layer1 self._make_layers(block, 64, layers[0], stride1)self.layer2 self._make_layers(block, 128, layers[1], stride2)self.layer3 self._make_layers(block, 256, layers[2], stride2)self.layer4 self._make_layers(block, 512, layers[3], stride2)self.avgpool nn.AdaptiveAvgPool2d((1, 1))self.fc nn.Linear(512 * block.expansion, num_classes)for m in self.modules():if isinstance(m, nn.Conv2d):n m.kernel_size[0] * m.kernel_size[1] * m.out_channelsm.weight.data.normal_(0, math.sqrt(2. / n))elif isinstance(m, nn.BatchNorm2d):m.weight.data.fill_(1)m.bias.data.zero_()def _make_layers(self, block, planes, num_blocks, stride1):downsample Noneif stride ! 1 or self.inplanes ! planes * block.expansion:downsample nn.Sequential(nn.Conv2d(self.inplanes, planes * block.expansion,kernel_size1, stridestride, biasFalse),nn.BatchNorm2d(planes * block.expansion),)layers []layers.append(block(self.inplanes, planes, stride, downsample))self.inplanes planes * block.expansionfor i in range(1, num_blocks):layers.append(block(self.inplanes, planes))return nn.Sequential(*layers)def forward(self, x):x self.conv1(x)x self.bn1(x)x self.relu(x)x self.maxpool(x)x self.layer1(x)x self.layer2(x)x self.layer3(x)x self.layer4(x)x self.avgpool(x)x x.view(x.size(0), -1)x self.fc(x)return xdef epsanet50():model EPSANet(EPSABlock, [3, 4, 6, 3], num_classes1000)return modeldef epsanet101():model EPSANet(EPSABlock, [3, 4, 23, 3], num_classes1000)return model
参考
EPSANet: 一种高效的多尺度通道注意力机制主要提出了金字塔注意力模块即插即用效果显著已开源
EPSANet金字塔拆分注意力模块
