search

HowieMa / PPT

[ECCV 2022] "PPT: token-Pruned Pose Transformer for monocular and multi-view human pose estimation"
55 stars 1 forks source link

About Transformer Block #5

Closed Berry-Wu closed 1 year ago

Berry-Wu commented 1 year ago

Hi, I have a question about the code in tokenpose_base.py of Transformer: why remove the first residual of Attention. I see the code in TokenPose dosen't remove it. And in your paper ,i can't find the reason.

HowieMa commented 1 year ago

Hi, thanks for your interest in our work. Actually, we still have the residual layer, as you can see from here. So, we don't need to talk about it in the paper.

Berry-Wu commented 1 year ago

Sorry, I missed your reply. It was my fault, thank you for your reply！:)