Real-Time Gaze Depth Estimation Using a Wearable Eye-Tracking Device

The accurate estimation of gaze depth is crucial for enhancing the functionality of eye-tracking systems, particularly in applications requiring precise three-dimensional gaze localization. This thesis presents a comprehensive approach to real-time gaze depth estimation using Tobii Pro Glasses 3, leveraging machine learning algorithms and precise pupil detection methods. The study begins with a theoretical overview of eye vergence and pupil detection techniques, followed by the development of a regression model trained to estimate gaze depth from the pupil positions. The experimental setup includes real-time video decoding and the analysis of pupil positions to validate the model’s performance. This research contributes to the field of gaze tracking by providing a robust framework for real-time depth estimation, highlighting the importance of precise pupil detection and positional calibration. Future work will focus on refining the model for broader applications and exploring additional features to enhance the accuracy and applicability of gaze depth estimation systems.