| Title |
Adversarial Domain Augmentation for Single Source Domain Generalization in Facial Expression Recognition |
| Authors |
김희제(Heeje Kim) ; 송병철(Byung Cheol Song) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.1.91 |
| Keywords |
Single-source domain generalization; Facial expression recognition; Adversarial attack; Domain augmentation |
| Abstract |
One of the major unresolved challenges in the field of facial expression recognition is performance degradation due to distribution shift between train and test domain. In this paper, we propose an adversarial domain augmentation method that applies perturbations to domain variant and invariant regions. The proposed method achieves up to 25% higher recognition accuracy compared to existing techniques, and also demonstrates its superiority through qualitative analysis. |