ResNet网络结构，BN以及迁移学习的详解

### ResNet50 架构详解 ResNet50 是一种深度卷积神经网络架构，拥有 50 层残差块。这种设计使得模型能够更有效地训练更深的网络层次，从而提高性能和鲁棒性[^1]。 #### 主要组成部分 - **输入层**：接受图像数据作为输入。 - **初始卷积层**：通常是一个 $7 \times 7$ 的大尺寸卷积核，步幅为 2，用于提取基础特征。 - **最大池化层**：紧随初始卷积之后，进一步减少空间维度。 - **四个阶段的残差模块**： - 每个阶段包含若干个残差单元（residual blocks）。这些单元内部采用短连接机制来缓解梯度消失问题。 - **全局平均池化层**：位于最后一个阶段之后，负责将特征图转换成固定长度向量表示。 - **全连接分类器**：最终输出类别预测概率分布。 ```python import torch.nn as nn class Bottleneck(nn.Module): expansion = 4 def __init__(self, inplanes, planes, stride=1, downsample=None): super(Bottleneck, self).__init__() self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False) self.bn1 = nn.BatchNorm2d(planes) self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=False) self.bn2 = nn.BatchNorm2d(planes) self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1, bias=False) self.bn3 = nn.BatchNorm2d(planes * self.expansion) self.relu = nn.ReLU(inplace=True) 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) out = self.relu(out) out = self.conv3(out) out = self.bn3(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out def make_layer(block, planes, blocks, stride=1): layers = [] downsample = None if stride != 1 or inplanes != block.expansion * planes: downsample = nn.Sequential( nn.Conv2d(inplanes, block.expansion*planes, kernel_size=1, stride=stride, bias=False), nn.BatchNorm2d(block.expansion*planes)) layers.append(block(inplanes, planes, stride, downsample)) inplanes = block.expansion * planes for i in range(1, blocks): layers.append(block(inplanes, planes)) return nn.Sequential(*layers) class ResNet(nn.Module): def __init__(self, block=Bottleneck, num_classes=1000): global inplanes inplanes = 64 super(ResNet, self).__init__() self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3,bias=False) self.bn1 = nn.BatchNorm2d(64) self.relu = nn.ReLU(inplace=True) self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) self.layer1 = make_layer(block, 64, 3) self.layer2 = make_layer(block, 128, 4, stride=2) self.layer3 = make_layer(block, 256, 6, stride=2) self.layer4 = make_layer(block, 512, 3, stride=2) self.avgpool = nn.AdaptiveAvgPool2d((1, 1)) self.fc = nn.Linear(512 * block.expansion, num_classes) # 初始化权重... def _forward_impl(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 = torch.flatten(x, 1) x = self.fc(x) return x model = ResNet() print(model) ``` ### 微调方法 对于特定任务的数据集较小的情况，可以通过迁移学习的方式利用预训练好的 ResNet50 来加速收敛并提升泛化能力： - 替换最后一层全连接层以适应新的分类数目； - 冻结前面大部分层的参数不变，只更新新增加部分以及少量最后几层的权值； - 使用较低的学习率重新训练整个网络； 具体实现如下所示： ```python from torchvision import models # 加载预训练模型 pretrained_model = models.resnet50(pretrained=True) for param in pretrained_model.parameters(): param.requires_grad_(False) # 冻结所有参数 num_ftrs = pretrained_model.fc.in_features pretrained_model.fc = nn.Linear(num_ftrs, new_num_classes) # 修改输出节点数 criterion = nn.CrossEntropyLoss() # 定义损失函数 optimizer = optim.Adam(filter(lambda p: p.requires_grad, pretrained_model.parameters()), lr=learning_rate) # 只优化可训练参数 ```