Contrastive Adversarial Learning for Person Independent Facial Emotion Recognition

Since most facial emotion recognition (FER) methods significantly rely on supervision information, they have a limit to analyzing emotions independently of persons. On the other hand, adversarial learning is well-known approach for generalized representation because it never requires information. This paper presents new FER. In detail, proposed enables FER network better understand complex emotional elements inherent in strong by adversarially weak samples based samples. As result, method can recognize persons understands expressions more accurately. addition, we propose contrastive loss function efficient learning. Finally, scheme was theoretically verified, and experimentally proven show state art (SOTA) performance.