COMETIC

2024
Chang Liu, Xiangyang Wang, Chun Yu, Yingtian Shi, Chongyang Wang, Ziqi Liu, Chen Liang, Yuanchun Shi

Limited accuracy of eye tracking on smartphones restricts its use. Existing RGB-camera-based eye tracking systems rely on extensive datasets, which could be improved by continuous fine-tuning using calibration data implicitly collected from user interactions. In this context, we propose COMETIC (Cursor Operation Mediated Eye-Tracking Implicit Calibration), which introduces a cursor-based interaction and utilizes the inherent correlation between cursor and eye movement.

Publication:

• Enhancing Smartphone Eye Tracking with Cursor-Based Interactive Implicit Calibration(CHI ’25).

Eye Tracking on Smartphone

Current smartphone eye tracking faces two main challenges: limited personalization for individual users and difficulty adapting to frequent posture changes. To address these issues, we developed a system that dynamically calibrates eye tracking through implicit user interactions, improving both accuracy and personalization.

COMETIC

By filtering valid cursor coordinates as proxies for gaze ground truth and fine-tuning the eye-tracking model with corresponding images, COMETIC enhances accuracy during interaction. Both filtering and fine-tuning utilize pre-trained models and can be further improved with personalized, dynamically updated data.

Data Collection Experiment

Evaluation

Results show that COMETIC achieves an average eye-tracking error of 208.04 px (1.2 cm), representing a 49.64% improvement compared to the system without fine-tuning. The analysis reveals that filtering cursor points with an actual gaze distance between 250 and 300 px (1.44 to 1.73 cm) yields the best eye-tracking results.