This paper presents an eye model that was constructed from a geometric simplification of the eye and anthropometric data about eye feature sizes in order to circumvent calibration per individual user. It was designed for efficiently performing uncalibrated eye gaze estimation on a user with frontal head pose. The model was implemented and tested on a database of still images obtained from 5 subjects. It was then compared against theoretical data. The model results in 3◦ of angular gaze estimation error which provides comparable performance to other methods without requiring calibration.
Date of creation, presentation, or exhibit
Department, Program, or Center
Computer Engineering (KGCOE)
Hnatow, J., Savakis, A., "An Efficient eye model for uncalibrated eye gaze enstimation." (2006)
RIT – Main Campus