chaneyddtt / Generating-Multiple-Hypotheses-for-3D-Human-Pose-Estimation-with-Mixture-Density-Network

Code for our CVPR2019 paper: Generating Multiple Hypotheses for 3D Human Pose Estimation with Mixture Density Network
MIT License
97 stars 9 forks source link

How to generate 3D human pose with our own data? #11

Open jellyfish1456 opened 3 years ago

jellyfish1456 commented 3 years ago

Hi there~ Thank you for sharing this great work.

My question is that how to generate 3D human pose with our own data? Thanks~

chaneyddtt commented 3 years ago

Hi @christw16, Can you give more details on your own data? I think you can directly train the model with your own data if the data includes 2d-to-3d joint correspondences.