LatentGaze: Cross-Domain Gaze Estimation through Gaze-Aware Analytic Latent Code Manipulation

TL;DR

LatentGaze introduces a novel cross-domain gaze estimation method that leverages latent code manipulation and GAN inversion to enhance gaze feature relevance and improve accuracy across datasets.

Contribution

The paper presents a gaze-aware latent code manipulation technique using GAN inversion to improve cross-domain gaze estimation performance.

Findings

Achieves state-of-the-art accuracy in cross-domain gaze estimation

Effectively disentangles gaze-relevant features from irrelevant ones

Utilizes gaze distortion loss to preserve gaze information

Abstract

Although recent gaze estimation methods lay great emphasis on attentively extracting gaze-relevant features from facial or eye images, how to define features that include gaze-relevant components has been ambiguous. This obscurity makes the model learn not only gaze-relevant features but also irrelevant ones. In particular, it is fatal for the cross-dataset performance. To overcome this challenging issue, we propose a gaze-aware analytic manipulation method, based on a data-driven approach with generative adversarial network inversion's disentanglement characteristics, to selectively utilize gaze-relevant features in a latent code. Furthermore, by utilizing GAN-based encoder-generator process, we shift the input image from the target domain to the source domain image, which a gaze estimator is sufficiently aware. In addition, we propose gaze distortion loss in the encoder that prevents the distortion of gaze information. The experimental results demonstrate that our method achieves state-of-the-art gaze estimation accuracy in a cross-domain gaze estimation tasks. This code is available at https://github.com/leeisack/LatentGaze/.