A new mapping function to improve the accuracy of a video-based eye tracker
Pieter J. Blignaut · 2013
The accuracy that can be achieved with a simple one camera, one IR source video-based eye tracker is evaluated for various combinations of calibration set and mapping model. It is shown that, using regression-based gaze estimation, accuracy varies across the display. A good mapping model would be one that results in a small average error as well as a small number of points with larger errors. An average error of less than 0.6° can be achieved when at least 12 calibration points are used in combination with a newly derived mapping model.