Machine Learning in Gaze-Based Interaction: A Survey of Eye Movements Events Detection

Spontaneous gaze-based input offers faster and more intuitive human-computer interaction as people naturally look at their desired destination. Developing a spontaneous gaze-based application faces many challenges, one of them is detecting events of eye movements. Most events detection methods are based on velocity or dispersion threshold. Unfortunately, these approaches depend on manual setting of threshold parameters. Despite previous attempts to review various techniques used in spontaneous gaze-based interaction, there is no survey paper that takes into account various machine learning techniques used for events detection. Here we present a brief overview of spontaneous gaze-based interactive applications and some machine learning approaches used for eye movement events detection. First, we explored major development of spontaneous gaze-based interaction by reviewing some examples of its applications. Next, we presented some state-of-the-art of threshold-based and machine learning-based events detection techniques. Finally, we discussed future research on spontaneous gaze-based interaction. Our brief survey paper maybe used by entry level researchers interested in developing uncalibrated interactive applications based on eye tracking.