#(Facial attribute estimation and expression recognition based on contextual channel attention mechanism)
A PyTorch implementation of the FAER.
We provide the training code for AffectNet and RAF-DB.
For AffectNet、RAF-DB dataset, run:
Comming soon
For CelebA dataset, run:
CUDA_VISIBLE_DEVICES=0,1 python main_celeba.py
Pre-trained models can be downloaded for evaluation as following:
task | accuracy | link |
---|---|---|
CelebA | 91.87% | Comming soon |
AffectNet | 66.66% | Comming soon |
RAF-DB | 91.75% | Comming soon |