Towards a General Deep Feature Extractor for Facial Expression Recognition

TL;DR

This paper introduces DeepFEVER, a deep learning-based facial feature extractor that generalizes well across multiple datasets for emotion recognition, outperforming existing methods and demonstrating strong cross-dataset applicability.

Contribution

The paper presents DeepFEVER, a novel deep neural network that learns a universal facial feature representation applicable to various emotion recognition datasets.

Findings

DeepFEVER outperforms state-of-the-art on AffectNet and Google datasets.

Features extracted by DeepFEVER generalize well to unseen datasets like RAF.

DeepFEVER demonstrates strong cross-dataset generalization capabilities.

Abstract

The human face conveys a significant amount of information. Through facial expressions, the face is able to communicate numerous sentiments without the need for verbalisation. Visual emotion recognition has been extensively studied. Recently several end-to-end trained deep neural networks have been proposed for this task. However, such models often lack generalisation ability across datasets. In this paper, we propose the Deep Facial Expression Vector ExtractoR (DeepFEVER), a new deep learning-based approach that learns a visual feature extractor general enough to be applied to any other facial emotion recognition task or dataset. DeepFEVER outperforms state-of-the-art results on the AffectNet and Google Facial Expression Comparison datasets. DeepFEVER's extracted features also generalise extremely well to other datasets -- even those unseen during training -- namely, the Real-World Affective Faces (RAF) dataset.