Related Experiment Video
Updated: Sep 6, 2025
Gaze in Action: Head-mounted Eye Tracking of Children's Dynamic Visual Attention During Naturalistic Behavior
Published on: November 14, 2018
High-Accuracy 3D Gaze Estimation with Efficient Recalibration for Head-Mounted Gaze Tracking Systems
Yang Xia1, Jiejunyi Liang1, Quanlin Li1
1State Key Laboratory of Digital Manufacturing Equipment and Technology, School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China.
This study introduces a new 3D gaze estimation method using head pose tracking to improve accuracy and reduce recalibration burden for head-mounted gaze trackers (HMGTs). The approach enhances user experience and achieves significant error reduction compared to existing methods.
Area of Science:
- Computer Vision
- Human-Computer Interaction
- Biomedical Engineering
Background:
- 3D gaze estimation using head-mounted gaze trackers (HMGTs) is challenging due to the human visual system's complexity.
- Current regression-based methods for gaze estimation suffer from fitting inaccuracies, extrapolation errors, and demanding recalibration procedures.
Purpose of the Study:
- To propose a high-accuracy 3D gaze estimation method with an efficient recalibration approach using head pose tracking.
- To address the limitations of existing methods, including inadequate fitting performance and degraded user experience.
Main Methods:
- Estimated key parameters (eyeball center, camera optical center) in the head frame using a geometry-based method.
- Developed a mapping relationship between direction features for visual axis calculation, utilizing accurately estimated parameters.
- Implemented a single-point recalibration method with an updated eyeball coordinate system.
Main Results:
- Achieved a 35% improvement in gaze estimation accuracy (mean error reduced from 2.00 to 1.31 degrees) through calibration.
- Demonstrated a 30% improvement in recalibration accuracy (mean error reduced from 2.00 to 1.41 degrees).
- Reduced the complexity of the mapping relationship and minimized extrapolation errors.
Conclusions:
- The proposed geometry-based 3D gaze estimation method significantly enhances accuracy and user experience.
- The efficient recalibration approach substantially reduces the burden of calibration procedures for HMGTs.
- This work offers a more robust and user-friendly solution for 3D gaze tracking applications.
