Behaviour4All: in-the-wild Facial Behaviour Analysis Toolkit

TL;DR

Behavior4All is an open-source, comprehensive toolkit for in-the-wild facial behavior analysis that integrates multiple tasks, leverages large datasets, and outperforms existing methods in accuracy, fairness, and speed.

Contribution

It introduces a novel framework combining distribution matching and label co-annotation to handle non-overlapping annotations in facial behavior analysis.

Findings

Outperforms state-of-the-art in accuracy and fairness

Demonstrates superior generalizability to unseen databases

Operates significantly faster than existing toolkits

Abstract

In this paper, we introduce Behavior4All, a comprehensive, open-source toolkit for in-the-wild facial behavior analysis, integrating Face Localization, Valence-Arousal Estimation, Basic Expression Recognition and Action Unit Detection, all within a single framework. Available in both CPU-only and GPU-accelerated versions, Behavior4All leverages 12 large-scale, in-the-wild datasets consisting of over 5 million images from diverse demographic groups. It introduces a novel framework that leverages distribution matching and label co-annotation to address tasks with non-overlapping annotations, encoding prior knowledge of their relatedness. In the largest study of its kind, Behavior4All outperforms both state-of-the-art and toolkits in overall performance as well as fairness across all databases and tasks. It also demonstrates superior generalizability on unseen databases and on compound expression recognition. Finally, Behavior4All is way times faster than other toolkits.