Towards Self-Supervised Gaze Estimation

TL;DR

This paper introduces SwAT, an equivariant self-supervised learning method based on SwAV, which improves gaze estimation accuracy by learning more informative representations from unlabeled face images, outperforming existing methods.

Contribution

The paper proposes SwAT, a novel equivariant self-supervised approach for gaze estimation that leverages geometric transformations to enhance representation learning.

Findings

SwAT outperforms state-of-the-art gaze estimation methods.

Achieves up to 57% improvement in cross-dataset evaluation.

Achieves up to 25% improvement in within-dataset evaluation.

Abstract

Recent joint embedding-based self-supervised methods have surpassed standard supervised approaches on various image recognition tasks such as image classification. These self-supervised methods aim at maximizing agreement between features extracted from two differently transformed views of the same image, which results in learning an invariant representation with respect to appearance and geometric image transformations. However, the effectiveness of these approaches remains unclear in the context of gaze estimation, a structured regression task that requires equivariance under geometric transformations (e.g., rotations, horizontal flip). In this work, we propose SwAT, an equivariant version of the online clustering-based self-supervised approach SwAV, to learn more informative representations for gaze estimation. We demonstrate that SwAT, with ResNet-50 and supported with uncurated unlabeled face images, outperforms state-of-the-art gaze estimation methods and supervised baselines in various experiments. In particular, we achieve up to 57% and 25% improvements in cross-dataset and within-dataset evaluation tasks on existing benchmarks (ETH-XGaze, Gaze360, and MPIIFaceGaze).