search

YuliangXiu / ECON

[CVPR'23, Highlight] ECON: Explicit Clothed humans Optimized via Normal integration
https://xiuyuliang.cn/econ
Other
1.06k stars 105 forks source link

Regarding the issue of mesh hand reconstruction #101

Closed xx674967 closed 8 months ago

xx674967 commented 8 months ago

Hi, thank you for releasing such an excellent work. During my testing, I have noticed that many image reconstruction results have errors in hand reconstruction around the waist, as shown in the following image. How can this issue be resolved? I look forward to your response.

econ_issue

YuliangXiu commented 8 months ago

It is not easy to fix such an issue, since the pocket regions are occluded in the given image. I think this could be alleviated under use_ifnet=False mode.

xx674967 commented 8 months ago

It is not easy to fix such an issue, since the pocket regions are occluded in the given image. I think this could be alleviated under use_ifnet=False mode.

Thank you for your response. I noticed that the TECH code has released. Can TECH handle this kind of case？ thanks ：）

YuliangXiu commented 8 months ago

It is not easy to fix such an issue, since the pocket regions are occluded in the given image. I think this could be alleviated under use_ifnet=False mode.

Thank you for your response. I noticed that the TECH code has released. Can TECH handle this kind of case？ thanks ：）

As there is a multi-pose training strategy in TeCH, such artifacts caused by occlusion could be reduced somehow.

xx674967 commented 8 months ago

It is not easy to fix such an issue, since the pocket regions are occluded in the given image. I think this could be alleviated under use_ifnet=False mode.

Thank you for your response. I noticed that the TECH code has released. Can TECH handle this kind of case？ thanks ：）

As there is a multi-pose training strategy in TeCH, such artifacts caused by occlusion could be reduced somehow.

Thanks, I will close this issue, have a nice day~