search

hongsukchoi / Pose2Mesh_RELEASE

Official Pytorch implementation of "Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Pose", ECCV 2020
MIT License
621 stars 69 forks source link

3d keypoints #10

Open foggrasshopperfish opened 3 years ago

foggrasshopperfish commented 3 years ago

Hi, I replaced the 3d keypoints with the ones that I got from VideoPose3D and the generated mesh look really weird. Is it because of any normalization that you processed with the 3D keypoints? Thanks.

hongsukchoi commented 3 years ago

Hi, I don't know the format of VideoPose3D's 3D keypoints output. But here are two possible reasons for your problem.

  1. Input joint order may be different. Check this part of README

  2. Yes, normalization of "2D keypoints" may be different. Check this code and the 2D and 3D poses' concatenation code.

If something's still unclear, you can debug the input and output of Pose2Mesh using import pdb; pdb.set_trace(). Remember to set num_workers of the dataloader as 0 if you do pdb in getitem function.

Thanks for your interest on my work :)