search

microsoft / multiview-human-pose-estimation-pytorch

This is an official Pytorch implementation of "Cross View Fusion for 3D Human Pose Estimation, ICCV 2019".
MIT License
541 stars 89 forks source link

How to do "direct triangulation" on baseline "Single" #34

Open Zhaoxuyu opened 3 years ago

Zhaoxuyu commented 3 years ago

I have a question on chaptar 6.3, table 2.

I want to know how you lifting 3D Pose from 2D Pose for baseline "Single". The paper says "direct triangulation".

According to my understanding, the input of "direct triangulation" is multiple view.

But for output of baseline "Single", it is only single view, how to do "direct triangulation" for it?

haibo-qiu commented 3 years ago

@Zhaoxuyu Yes, the "direct triangulation" takes multi-view images as input.

The H36M provides four views, but the baseline "Single" uses separated view images for training without any fusion. Then we utilize the trained "Single" model to obtain the heatmaps of four views images, which will be taken by "Triangulation" to get the 3D pose.

99sun99 commented 2 years ago

So you didn't use the lib/multiview/triangulate.py in your training process? why? I think you write it for a reason~