Multi-modal Facial Action Unit Detection with Large Pre-trained Models for the 5th Competition on Affective Behavior Analysis in-the-wild

TL;DR

This paper introduces a multi-modal approach for facial action unit detection using large pre-trained models across visual, acoustic, and lexical features, achieving competitive results in the ABAW 2023 Challenge.

Contribution

It presents a novel multi-modal method combining visual, acoustic, and lexical features with pre-trained models for improved AU detection.

Findings

Achieved an F1 score of 52.3% on ABAW 2023 validation set.

Enhanced visual features with super-resolution and face alignment.

Demonstrated the effectiveness of multi-modal features in AU detection.

Abstract

Facial action unit detection has emerged as an important task within facial expression analysis, aimed at detecting specific pre-defined, objective facial expressions, such as lip tightening and cheek raising. This paper presents our submission to the Affective Behavior Analysis in-the-wild (ABAW) 2023 Competition for AU detection. We propose a multi-modal method for facial action unit detection with visual, acoustic, and lexical features extracted from the large pre-trained models. To provide high-quality details for visual feature extraction, we apply super-resolution and face alignment to the training data and show potential performance gain. Our approach achieves the F1 score of 52.3% on the official validation set of the 5th ABAW Challenge.