Boosting Facial Action Unit Detection Through Jointly Learning Facial Landmark Detection and Domain Separation and Reconstruction
This addresses the challenge of improving AU detection accuracy in real-world, unconstrained environments for applications like emotion analysis, though it appears incremental as it builds on existing multi-task and contrastive learning methods.
The paper tackled the problem of incorporating unlabeled facial images into supervised Facial Action Unit (AU) detection frameworks by proposing a multi-task learning approach that jointly learns AU domain separation and reconstruction with facial landmark detection, achieving state-of-the-art results on two benchmarks.
Recently how to introduce large amounts of unlabeled facial images in the wild into supervised Facial Action Unit (AU) detection frameworks has become a challenging problem. In this paper, we propose a new AU detection framework where multi-task learning is introduced to jointly learn AU domain separation and reconstruction and facial landmark detection by sharing the parameters of homostructural facial extraction modules. In addition, we propose a new feature alignment scheme based on contrastive learning by simple projectors and an improved contrastive loss, which adds four additional intermediate supervisors to promote the feature reconstruction process. Experimental results on two benchmarks demonstrate our superiority against the state-of-the-art methods for AU detection in the wild.