search

karfly / learnable-triangulation-pytorch

This repository is an official PyTorch implementation of the paper "Learnable Triangulation of Human Pose" (ICCV 2019, oral). Proposed method archives state-of-the-art results in multi-view 3D human pose estimation!
MIT License
1.1k stars 181 forks source link

evalution results in human3.6M? #44

Closed ericzw closed 4 years ago

ericzw commented 4 years ago

Hi,

I saw that you use every fifth frame for evaluation in your paper. Is it means that the MPJPE of 20.8 is evaluated on every fifth frame?

Have you evaluated your model on the full frames? Could you provide the results on the full frames? Thanks

karfly commented 4 years ago

Hi, @qiuzhongwei-USTB.

We found out that it doesn’t affect the resulting accuracy a lot.

Before the commit #42 in this repository was done on all frames resulting in 20.5 mm MPJPE error. Honestly, the weights of the pretrained model we uploaded to the repository were not the ones we used in the paper (current model was trained for a longer time). That’s why the error is lower.