GridDehazeNet

GridDehazeNet是一个端到端可训练的卷积神经网络，在单幅图像去雾方面具有竞争性能。它由三个模块组成：预处理、主干和后处理。预处理模块产生具有更好多样性和更相关特征的学习输入。主干模块在网格网络上实现了一种基于注意力的多尺度估计，可以有效缓解传统多尺度方法中的瓶颈问题。后处理模块有助于减少最终输出中的伪像。GridDehazeNet在合成图像和真实图像上都表现优于当前最先进的方法。此外，GridDehazeNet还具有灵活性，可以通过选择合适的注意力权重来修剪或停用部分网络，或者移除交换分支以获得类似于传统多级网络的结构。<span class="em">1</span><span class="em">2</span><span class="em">3</span>
#### 引用[.reference_title]
- *1* [论文阅读：GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing](https://blog.csdn.net/space_walk/article/details/108085608)[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^v93^chatsearchT3_2"}}] [.reference_item style="max-width: 50%"]
- *2* *3* [GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing(ICCV2019)](https://blog.csdn.net/change_lkl/article/details/127159524)[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^v93^chatsearchT3_2"}}] [.reference_item style="max-width: 50%"]
[ .reference_list ]