Unofficial implementation of paper: Estimation of continuous valence and arousal levels from faces in naturalistic conditions
Code is partly forked/copied from the official code of emonet
python train.py --nclassses 5
python train.py --nclassses 5 --kd --kd_w 0.3 --path step1_model_path
https://github.com/jingyang2017/emonet_test/tree/master/ibug/emotion_recognition/emonet/weights
[1] Toisoul, Antoine, Jean Kossaifi, Adrian Bulat, Georgios Tzimiropoulos, and Maja Pantic. "Estimation of continuous valence and arousal levels from faces in naturalistic conditions." Nature Machine Intelligence 3, no. 1 (2021): 42-50.
Apply emonet_train to facial action unit detection: https://github.com/jingyang2017/aunet_train