search

postech-ami / FastMETRO

[ECCV'22] Official PyTorch Implementation of "Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers"
https://fastmetro.github.io/
MIT License
164 stars 14 forks source link

About edge length loss (GT supervision) & normal vector loss #7

Closed imabackstabber closed 1 year ago

imabackstabber commented 1 year ago

They didn't seems to be appeared in your paper. Is there any existing evaluation result when training using just 2d and 3d loss?

FastMETRO commented 1 year ago

Hello,

We recently added the edge length loss and normal vector loss for visually pleasing mesh results. As reported in Table 14 of Pose2Mesh, those loss functions have nearly no effect on quantitative results.

Note that the evaluation results in our paper were produced without using those loss functions.

Thanks for your interest in our work!!

FastMETRO commented 1 year ago

Please reopen this issue if you need more help regarding this.