seresnet18相比resnet18

### SeresNet18 和 ResNet18 的架构性能对比

#### 架构设计上的区别
ResNet18 是一种基础的残差网络结构，由多个两层卷积模块组成。每个模块通过跳跃连接实现信息传递，从而缓解深层神经网络中的梯度消失问题[^3]。

SeresNet18 则是在 ResNet18 基础上引入了 SE（Squeeze-and-Excitation）机制。SE 模块能够自适应地重新校准通道特征响应，在不显著增加计算成本的情况下提升模型表达能力。具体来说，SE 结构会先全局池化获取各个通道的重要性权重，再利用这些权重重加权原始特征图，使得重要特征得到增强而次要特征被抑制[^1]。

#### 性能表现的不同
由于加入了注意力机制，SeresNet18 对于不同类别间细微差异的学习更加敏感，尤其在处理复杂背景下的目标识别任务时表现出更好的鲁棒性和准确性。实验表明，在相同条件下训练相同的 epoch 数量之后，通常情况下 SeresNet18 能够取得优于标准版 ResNet18 的分类精度[^4]。

import torch.nn as nn

class BasicBlock(nn.Module):
    expansion = 1
    
    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(BasicBlock, self).__init__()
        
        # Standard Convolution Layers for both networks
        self.conv1 = conv3x3(inplanes, planes, stride)
        self.bn1 = nn.BatchNorm2d(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = conv3x3(planes, planes)
        self.bn2 = nn.BatchNorm2d(planes)

        # Squeeze and Excitation layer only present in SEResNet
        self.se_layer = SELayer(planes * self.expansion)

        self.downsample = downsample
        self.stride = stride
        
    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        # Apply squeeze-excitation mechanism if available
        if hasattr(self, 'se_layer'):
            out = self.se_layer(out)

        if self.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out


def se_resnet18(pretrained=False, **kwargs):
    """Constructs a SeResNet-18 model."""
    model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
    
    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),
                nn.ReLU(inplace=True),
                nn.Linear(channel // reduction, channel),
                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)


def resnet18(pretrained=False, **kwargs):
    """Constructs a ResNet-18 model."""
    model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)