Gaze Zone Estimation Using Head Pose Angle and Iris Center Localization on Simulation Environment

Abstract

Eye gaze estimation has importance role in various computer vision algorithm such as driver monitoring, non-contact human-computer interaction, etc. Various gaze zone estimation have been developed by various researchers and engineers. In this paper, we will show a coarse gaze estimation by using a single RGB camera that implemented on two computer screens. We divided the zone into 8 zones with the same size for each zone. We utilized head pose and iris center as the feature to train a model for estimating gaze zone. To classify the gaze zone, we used support vector machine (SVM) with radial basis function (RBF) kernel. We made two experiment which one experiment only used head pose feature only and the other added with the iris center coordinate. The result shows that by using head pose and iris center coordinate on gaze zone estimation has slightly higher than the head pose-alone experiment. Our proposed method also showed promising result which has accuracy higher than 91%.