resnet18网络结构简图
时间: 2023-08-02 15:11:26 浏览: 119
回答: resnet18网络结构简图如下所示:
```
输入 -> 7x7卷积层,64个卷积核,步长2,填充3 -> 批量归一化 -> ReLU激活函数 -> 最大池化层,3x3池化核,步长2,填充1 -> resnet block1 -> resnet block2 -> resnet block3 -> resnet block4 -> 平均池化层,7x7池化核,步长1 -> 全连接层,输出类别数
```
其中,每个resnet block包含两个basic block,每个basic block的结构如下:
```
输入 -> 3x3卷积层,64个卷积核,步长1,填充1 -> 批量归一化 -> ReLU激活函数 -> 3x3卷积层,64个卷积核,步长1,填充1 -> 批量归一化 -> 残差连接 -> ReLU激活函数
```
在resnet block之间的残差连接使用1x1卷积层,而在resnet block内部的残差连接使用恒等映射(identity mapping)。\[1\]
#### 引用[.reference_title]
- *1* [Resnet-18网络图示理解](https://blog.csdn.net/weixin_42899627/article/details/108228782)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insertT0,239^v3^insert_chatgpt"}} ] [.reference_item]
- *2* [resnet50 网络结构全图](https://blog.csdn.net/dongtuoc/article/details/125467809)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insertT0,239^v3^insert_chatgpt"}} ] [.reference_item]
[ .reference_list ]
阅读全文