Representation Learning and Identity Adversarial Training for Facial Behavior Understanding

TL;DR

This paper introduces Face9M, a large facial image dataset, and proposes a novel identity adversarial training method combined with a masked autoencoder to improve facial action unit detection, achieving state-of-the-art results.

Contribution

The paper presents Face9M dataset, a new masked autoencoder model, and a novel identity adversarial training approach for more accurate and identity-invariant facial behavior understanding.

Findings

Face9M dataset improves AU detection performance.

IAT regularization reduces identity bias in models.

FMAE-IAT achieves state-of-the-art F1 scores on multiple datasets.

Abstract

Facial Action Unit (AU) detection has gained significant attention as it enables the breakdown of complex facial expressions into individual muscle movements. In this paper, we revisit two fundamental factors in AU detection: diverse and large-scale data and subject identity regularization. Motivated by recent advances in foundation models, we highlight the importance of data and introduce Face9M, a diverse dataset comprising 9 million facial images from multiple public sources. Pretraining a masked autoencoder on Face9M yields strong performance in AU detection and facial expression tasks. More importantly, we emphasize that the Identity Adversarial Training (IAT) has not been well explored in AU tasks. To fill this gap, we first show that subject identity in AU datasets creates shortcut learning for the model and leads to sub-optimal solutions to AU predictions. Secondly, we demonstrate that strong IAT regularization is necessary to learn identity-invariant features. Finally, we elucidate the design space of IAT and empirically show that IAT circumvents the identity-based shortcut learning and results in a better solution. Our proposed methods, Facial Masked Autoencoder (FMAE) and IAT, are simple, generic and effective. Remarkably, the proposed FMAE-IAT approach achieves new state-of-the-art F1 scores on BP4D (67.1\%), BP4D+ (66.8\%), and DISFA (70.1\%) databases, significantly outperforming previous work. We release the code and model at https://github.com/forever208/FMAE-IAT.