Mutual Information Regularized Identity-aware Facial ExpressionRecognition in Compressed Video

TL;DR

This paper introduces a novel method for facial expression recognition in compressed videos that uses mutual information minimization to produce identity-invariant expression features, enabling faster and effective FER without needing raw images.

Contribution

It proposes a mutual information regularization framework in the compressed domain that eliminates the need for identity labels and improves FER efficiency and accuracy.

Findings

Achieves comparable or better performance than image-based methods.

Runs approximately three times faster during inference.

Effectively disentangles expression features from identity factors.

Abstract

How to extract effective expression representations that invariant to the identity-specific attributes is a long-lasting problem for facial expression recognition (FER). Most of the previous methods process the RGB images of a sequence, while we argue that the off-the-shelf and valuable expression-related muscle movement is already embedded in the compression format. In this paper, we target to explore the inter-subject variations eliminated facial expression representation in the compressed video domain. In the up to two orders of magnitude compressed domain, we can explicitly infer the expression from the residual frames and possibly extract identity factors from the I frame with a pre-trained face recognition network. By enforcing the marginal independence of them, the expression feature is expected to be purer for the expression and be robust to identity shifts. Specifically, we propose a novel collaborative min-min game for mutual information (MI) minimization in latent space. We do not need the identity label or multiple expression samples from the same person for identity elimination. Moreover, when the apex frame is annotated in the dataset, the complementary constraint can be further added to regularize the feature-level game. In testing, only the compressed residual frames are required to achieve expression prediction. Our solution can achieve comparable or better performance than the recent decoded image-based methods on the typical FER benchmarks with about 3 times faster inference.