Reproduce “Evaluation of Appearance-Based Methods and Implications for Gaze-Based Applications” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Xucong Zhang, Yusuke Sugano, Andreas Bulling

This paper evaluates appearance-based gaze estimation methods, comparing their accuracy and robustness against model-based approaches and commercial eye trackers across diverse interaction conditions. It also introduces OpenGaze, an open-source software toolkit designed to facilitate the integration of these camera-based techniques into human-computer interaction applications.

Key insights

  • Appearance-based gaze estimation offers a robust alternative to specialized hardware, maintaining consistent accuracy (2.3-3.1 degrees error) across a wide operational distance (30-180 cm).
  • These methods can achieve accuracy competitive with commercial trackers with minimal personal calibration, requiring as few as four samples to reach approximately 2.5 degrees error.
  • Democratizing access to advanced gaze estimation techniques through user-friendly toolkits is crucial for expanding the scope and accessibility of gaze as an interaction modality in HCI.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.