How to visualize the realtime recognition in 3d format？

open-mmlab / mmskeleton

A OpenMMLAB toolbox for human pose estimation, skeleton-based action recognition, and action synthesis.

Apache License 2.0

2.9k stars 1.03k forks source link

How to visualize the realtime recognition in 3d format？ #301

Open antoniolq opened 4 years ago

antoniolq commented 4 years ago

hello, I want to visualize the realtime recognition in 3d format by using 3d pose estimation. But most of algorithms about 3d pose estimation output 17 key points? And the NTU-RGB+D has 25 key points? do you know how to make it? is that possible?

antoniolq commented 4 years ago

look forward to your reply!

a726 commented 4 years ago

您好，我想通过使用3d姿态估计来可视化3d格式的实时识别。但是大多数有关3d姿态估计的算法会输出17个关键点？NTU-RGB + D有25个关键点？你知道怎么做吗？那可能吗？

请问您实现了没有？期待您的回复！