Appearance-Based Gaze Tracking: A Brief Review

Human gaze tracking plays an important role in the field of Human-Computer Interaction. This paper presents a brief review on appearance-based gaze tracking. Based on the appearance of human eyes, input features can be classified into three categories according to the different ways of extracting human eyes features, namely, complete human eye image, pixel-based feature and 3D reconstruction image. The estimation process from human eye feature to fixation point mainly uses different mapping functions. In this paper, common mapping functions and related algorithms are described in detail: k-nearest neighbor (KNN), random forest (RF) regression, gaussian process (GP) regression, support vector machines (SVM) and artificial neural networks (ANN). This paper evaluates the performance of these gaze tracking algorithms using different mapping functions. Based on the results of the evaluation, potential challenges are summarized and the future directions of gaze estimation are prospected.