光流transformer

光流Transformer是一种基于Transformer架构的神经网络，用于光流估计。它被称为FlowFormer。FlowFormer的核心思想是将图像对转换为4D成本体积，并使用交替组转换器(AGT)层将成本编码到一个新的潜在空间中。然后，通过一个带有动态位置成本查询的循环Transform解码器对位置成本查询进行解码。FlowFormer采用了一种新颖的代价体编码器和循环代价解码器，通过迭代地改进估计的光流，实现了最先进的流估计性能。这项研究已经被ECCV 2022收录，并且作者也在Github上开源了该项目的代码。<span class="em">1</span><span class="em">2</span><span class="em">3</span>
#### 引用[.reference_title]
- *1* *2* [经典文献阅读之--FlowFormer(Transformer结构光流估计)](https://blog.csdn.net/lovely_yoshino/article/details/127344116)[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%"]
- *3* [【论文简述】FlowFormer：A Transformer Architecture for Optical Flow（ECCV 2022）](https://blog.csdn.net/qq_43307074/article/details/128652133)[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 ]