search

YuliangXiu / ICON

[CVPR'22] ICON: Implicit Clothed humans Obtained from Normals
https://icon.is.tue.mpg.de
Other
1.61k stars 220 forks source link

Questions about ICON+Keypoint #231

Open Jarvisss opened 1 year ago

Jarvisss commented 1 year ago

https://github.com/YuliangXiu/ICON/blob/8919d96754d106e19ee8ac9dcfaa1bd2b598764c/lib/net/HGPIFuNet.py#L301-L305 I find the implementation of keypoint+ICON projects query points to camera space, but smpl joints are not projected. But in original keypointNerf paper, the spatial encoding is calculated by the projected keypoints. https://github.com/YuliangXiu/ICON/blob/8919d96754d106e19ee8ac9dcfaa1bd2b598764c/lib/net/HGPIFuNet.py#L333-L340 And SMPL features are still used in the keypoint version, is this what it supposed to be?

Jarvisss commented 1 year ago

@YuliangXiu

YuliangXiu commented 1 year ago

smpl_joints should be projected into camera space, you could visualize the keypoints when running infer.py