openpose算法
时间: 2023-09-13 12:13:29 浏览: 172
openpose源码
OpenPose算法是一种流行的人体姿态评估算法,它通过检测图像中的关节(关键点)来实现。OpenPose的架构包括两个分支,Branch1和Branch2。Branch1使用CNN网络提取部位置信度图,而Branch2使用CNN网络提取部位关联度图(Part Affinity Fields)。通过将这两个信息结合,OpenPose利用图论中的偶匹配(Bipartite Matching)算法来连接同一个人的关节点,并生成完整的人体骨架。最后,基于关联度图(Part Affinity Fields),OpenPose使用匈牙利算法(Hungarian Algorithm)将多人姿态估计问题转换为图问题,进一步优化结果。<span class="em">1</span><span class="em">2</span><span class="em">3</span>
#### 引用[.reference_title]
- *1* [人体姿态估计算法之open pose](https://blog.csdn.net/qq_43258953/article/details/104441286)[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^v92^chatsearchT0_1"}}] [.reference_item style="max-width: 50%"]
- *2* *3* [openpose tensorflow版本 原理+项目代码详解](https://blog.csdn.net/surserrr/article/details/90371238)[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^v92^chatsearchT0_1"}}] [.reference_item style="max-width: 50%"]
[ .reference_list ]
阅读全文