search

MSA-LMC / S2D

[TAFFC 2024] The official implementation of paper: From Static to Dynamic: Adapting Landmark-Aware Image Models for Facial Expression Recognition in Videos
https://ieeexplore.ieee.org/document/10663980
Apache License 2.0
50 stars 1 forks source link

SFER model #11

Open s25813s5 opened 15 hours ago

s25813s5 commented 15 hours ago

HI, Could you provide the SFER model for Pretrain-AffectNet-7.pth?

cyinen commented 12 hours ago

You can found the model from this link

s25813s5 commented 8 hours ago

Sorry, I meant the ViT model architecture trained on AffectNet. I want to reproduce the performance of Pretrain-AffectNet-7.pth on the AffectNet dataset.

cyinen commented 7 hours ago

You can obtain the SFER model by removing the temporal module.

s25813s5 commented 6 hours ago

The classifier part seems to be different as well. Could you offer the classification parts of head and weak_head?