网站建设中搭建页面结构,电子商务公司门头照片,hemi网站怎么做热图,做擦边球视频网站文章目录 代码实现参考 代码实现
本文实现 ResNet原论文 Deep Residual Learning for Image Recognition 中的50层#xff0c;101层和152层残差连接。 代码中使用基础残差块这个概念#xff0c;这里的基础残差块指的是上图中红色矩形圈出的内容#xff1a;从上到下分别使用… 文章目录 代码实现参考 代码实现
本文实现 ResNet原论文 Deep Residual Learning for Image Recognition 中的50层101层和152层残差连接。 代码中使用基础残差块这个概念这里的基础残差块指的是上图中红色矩形圈出的内容从上到下分别使用3, 4, 6, 3个基础残差块每个基础残差块由三个卷积层组成核大小分别为1x1, 3x3, 1x1 。
残差连接的结构 复现代码如下
import torch
import torch.nn as nn# 基础残差块后面ResNet要多次重复使用该块
class block(nn.Module):def __init__(self, in_channels, out_channels, identity_downsampleNone, stride1):super(block, self).__init__()self.expansion 4 self.conv1 nn.Conv2d(in_channels, out_channels, kernel_size1, stride1, padding0)self.bn1 nn.BatchNorm2d(out_channels)self.conv2 nn.Conv2d(out_channels, out_channels, kernel_size3, stridestride, padding1)self.bn2 nn.BatchNorm2d(out_channels)self.conv3 nn.Conv2d(out_channels, out_channels*self.expansion, kernel_size1, stride1, padding0)self.bn3 nn.BatchNorm2d(out_channels*self.expansion)self.relu nn.ReLU()self.identity_downsample identity_downsampledef forward(self, x):identity xx self.conv1(x)x self.bn1(x)x self.relu(x)x self.conv2(x)x self.bn2(x)x self.relu(x)x self.conv3(x)x self.bn3(x)if self.identity_downsample is not None:identity self.identity_downsample(identity)x identityx self.relu(x)return xclass ResNet(nn.Module):def __init__(self, block, layers, image_channels, num_classes):super(ResNet, self).__init__()# 初始化的层self.in_channels 64self.conv1 nn.Conv2d(image_channels, 64, kernel_size7, stride2, padding3)self.bn1 nn.BatchNorm2d(64)self.relu nn.ReLU()self.maxpool nn.MaxPool2d(kernel_size3, stride2, padding1)# ResNet layersself.layer1 self._make_layer(block, layers[0], out_channels64, stride1)self.layer2 self._make_layer(block, layers[1], out_channels128, stride2)self.layer3 self._make_layer(block, layers[2], out_channels256, stride2)self.layer4 self._make_layer(block, layers[3], out_channels512, stride2)self.avgpool nn.AdaptiveAvgPool2d((1, 1))self.fc nn.Linear(512*4, num_classes)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.reshape(x.shape[0], -1)x self.fc(x)return x# 核心函数调用block基础残差块构造ResNet的每一层def _make_layer(self, block, num_residual_blocks, out_channels, stride):identity_downsample Nonelayers []if stride ! 1 or self.in_channels ! out_channels * 4:identity_downsample nn.Sequential(nn.Conv2d(self.in_channels, out_channels*4, kernel_size1,stridestride), nn.BatchNorm2d(out_channels*4))layers.append(block(self.in_channels, out_channels, identity_downsample, stride))self.in_channels out_channels * 4for i in range(num_residual_blocks - 1):layers.append(block(self.in_channels, out_channels)) # 256 - 64, 64*4(256) againreturn nn.Sequential(*layers)# 构造ResNet50层默认图像通道3分类类别为1000
def resnet50(img_channels3, num_classes1000):return ResNet(block, [3, 4, 6, 3], img_channels, num_classes)# 构造ResNet101层
def resnet101(img_channels3, num_classes1000):return ResNet(block, [3, 4, 23, 3], img_channels, num_classes)# 构造ResNet152层
def resnet152(img_channels3, num_classes1000):return ResNet(block, [3, 8, 36, 3], img_channels, num_classes)# 测试输出y的形状是否满足1000类
def test():net resnet152()x torch.randn(2, 3, 224, 224)y net(x)print(y.shape) # [2, 1000]test()参考
[1] Deep Residual Learning for Image Recognition [2] https://www.youtube.com/watch?vDkNIBBBvcPslistPLhhyoLH6IjfxeoooqP9rhU3HJIAVAJ3Vzindex19
