Random Forest Regression for continuous affect using Facial Action Units

TL;DR

This paper presents a method using facial Action Units and Random Forest regression to predict continuous affective states, evaluated on the ABAW in-the-wild dataset.

Contribution

It introduces a Random Forest regression approach with facial Action Units for continuous affect prediction in unconstrained environments.

Findings

Performed comparably to baseline methods

Utilized OpenFace for feature extraction

Applied to the ABAW in-the-wild dataset

Abstract

In this paper we describe our approach to the arousal and valence track of the 3rd Workshop and Competition on Affective Behavior Analysis in-the-wild (ABAW). We extracted facial features using OpenFace and used them to train a multiple output random forest regressor. Our approach performed comparable to the baseline approach.