Motivation of computing depth normal for ITOP and K2HPD datasets

zhangboshen / A2J

Code for paper "A2J: Anchor-to-Joint Regression Network for 3D Articulated Pose Estimation from a Single Depth Image". ICCV2019

MIT License

287 stars 46 forks source link

Motivation of computing depth normal for ITOP and K2HPD datasets #43

Closed bearpaw closed 2 years ago

bearpaw commented 3 years ago

Hi @zhangboshen , I noticed that you precomputed the depth normal for human pose estimation tasks, while you use original depth maps for hand pose estimation. What's the motivation for this? Does original depth not work for human pose estimation? Thanks!

zhangboshen commented 3 years ago

Hi @bearpaw , Yes we did precomputed depth normal for ITOP dataset, the reason that we discard normal is that depth map itself can already achieve promising results.