VIP: A unifying framework for computational eye-gaze research

Abstract

Eye-gaze is an emerging modality in many research areas and applications. We present our VIP framework, which captures the dependence of eye-gaze on Visual stimulus, Intent, and Person. The unifying framework characterizes current eye-gaze computational models. It allows computer scientists to formally define their research problems and to compare with other work. We review the state-of-art computational eye-gaze research and applications with reference to our framework. With the framework, we identify gaps in eye-gaze research and present our work on the new research problem of attribute classification. The accuracy of 0.92 is achieved for classification of Introvert/Extrovert. © 2013 Springer International Publishing.