Tracking Eyes and Monitoring Eye Gaze

In this paper, we present a non-intrusive eye tracker which can detect and track a user's eyes in real time as soon as the face appears in the view of the camera without special lighting or any marks on the user's face. We also discuss the problem of gaze monitoring. We employ neural networks to estimate a user's eye gaze using the images of user's both eyes as input to the neural networks. We have collected 4 sets of data from 4 different users using the eye tracker and have trained and tested several neural networks. The eye gaze monitoring system has achieved accuracy of between 1.3 and 1.8 degrees with a user dependent setup and of 1.9 degrees for a multi-user setup. 1. INTRODUCTION Focus of attention plays an important role in user modeling. In many applications, it required to monitor user's eye gaze to find user's focus of attention. Current eye gaze tracking methods basically rely on intrusive techniques such as measuring the reflection of some light (ususally infrared light) ..