Recognizing Cognitive Activities Through Eye Tracking

Abstract

Eye detection and tracking is usually performed by using specific devices that allow to determine the pupil position in many different situations. We propose to use these techniques for recognizing cognitive activities that a potential user is carrying out in front of a computer. We use the images captured by a conventional web camera located over the computer display. Those image are processed and, after the face and facial landmarks are found, the user gaze is analyzed and the ethogram and several statistics associated to the eyes and gaze destination are computed. They are used for determining what is doing the user from a set of predefined activities.