/FDRL

Unofficial implementation of "Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition - CVPR'21"

Primary LanguagePython

FDRL

Unofficial implementation of Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition - CVPR'21

Training

  1. First download the resnet18 pretrained on MS-Celeb-1M and put it in the backbone folder (please create it by yourself)
  2. To train the model according to the paper use the following command:
python train.py --exp_name {name_for_exp}

Testing

Use the following command:

python eval.py --resume {path_to_trained_model_in_checkpoints}

Note that if you had used other arguments at training, please make sure to apply them at testing.

Cite

@InProceedings{Ruan_2021_CVPR,
    author    = {Ruan, Delian and Yan, Yan and Lai, Shenqi and Chai, Zhenhua and Shen, Chunhua and Wang, Hanzi},
    title     = {Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2021},
    pages     = {7660-7669}
}

If you find this repo useful, please star⭐, tks.