在definedet网络的pooling层后加入SENet模块然后通过conv与deconv特征融合代码

以下是在Definedet网络的pooling层后加入SENet模块，然后通过conv与deconv特征融合的代码示例：

import torch.nn as nn
import torch.nn.functional as F

class DefinedetSENet(nn.Module):
    def __init__(self, num_classes=80):
        super(DefinedetSENet, self).__init__()
        self.base = DefinedetBase()
        self.pool = nn.AdaptiveAvgPool2d((1, 1))
        self.se = SELayer(2048, reduction=16)
        self.conv1 = nn.Conv2d(2048, 256, kernel_size=1, stride=1, padding=0)
        self.conv2 = nn.Conv2d(1024, 256, kernel_size=1, stride=1, padding=0)
        self.deconv1 = nn.ConvTranspose2d(256, 256, kernel_size=2, stride=2, padding=0)
        self.deconv2 = nn.ConvTranspose2d(256, 256, kernel_size=2, stride=2, padding=0)
        self.cls_head = nn.Conv2d(256, num_classes, kernel_size=3, stride=1, padding=1)
        self.reg_head = nn.Conv2d(256, 4, kernel_size=3, stride=1, padding=1)

    def forward(self, x):
        c3, c4, c5 = self.base(x)
        p5 = self.pool(c5)
        p5 = self.se(p5)
        p5 = self.conv1(p5)
        p4 = self.conv2(c4)
        p4 = p4 + self.deconv1(p5)
        p3 = c3
        p3 = p3 + self.deconv2(p4)
        out = F.relu(p3)
        out_cls = self.cls_head(out)
        out_reg = self.reg_head(out)
        return out_cls, out_reg

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):
        b, c, _, _ = x.size()
        y = self.avg_pool(x).view(b, c)
        y = self.fc(y).view(b, c, 1, 1)
        return x * y.expand_as(x)

在这个示例中，我们先定义了一个基础的Definedet网络，然后在其pooling层后加入了SENet模块，将特征图的channel数从2048减小到256。然后我们通过conv与deconv特征融合，将p5的特征图通过conv1压缩到256维，并与c4的特征图通过conv2压缩到256维后相加得到p4的特征图。然后将p4的特征图通过deconv1上采样2倍后与c3的特征图相加得到p3的特征图。最后将p3的特征图送入分类和回归头进行预测。