Privileged Attribution Constrained Deep Networks for Facial Expression Recognition

TL;DR

This paper introduces Privileged Attribution Loss (PAL), a novel method guiding deep networks to focus on key facial regions for improved facial expression recognition, outperforming state-of-the-art methods.

Contribution

The paper proposes PAL, a new attribution-guided loss that enhances model focus on salient facial areas without extra semantic info at test time.

Findings

PAL outperforms existing methods on RAF-DB and AffectNet datasets.

The method improves generalization on small and noisy datasets.

It is architecture-independent and does not require additional semantic data.

Abstract

Facial Expression Recognition (FER) is crucial in many research domains because it enables machines to better understand human behaviours. FER methods face the problems of relatively small datasets and noisy data that don't allow classical networks to generalize well. To alleviate these issues, we guide the model to concentrate on specific facial areas like the eyes, the mouth or the eyebrows, which we argue are decisive to recognise facial expressions. We propose the Privileged Attribution Loss (PAL), a method that directs the attention of the model towards the most salient facial regions by encouraging its attribution maps to correspond to a heatmap formed by facial landmarks. Furthermore, we introduce several channel strategies that allow the model to have more degrees of freedom. The proposed method is independent of the backbone architecture and doesn't need additional semantic information at test time. Finally, experimental results show that the proposed PAL method outperforms current state-of-the-art methods on both RAF-DB and AffectNet.