Eye-based Activity Recognition
Professor Hans Gellersen, School of Computing and Communications
Tuesday 14 December 2010, 1450-1515
Lecture Theatre 3, Management School Building
If only computers knew what we do, they might support us better (or not get as much in the way). Activity recognition is a research field in computer science that investigates sensor systems and computing methods for automated inferences about human activity. Early work in this field has developed methods that classify body movements and locomotion modes from analysis of inertial sensors worn by the user; in turn, this data can be used to reason about what the user is doing. This work has application in many domains, from assisted living where the focus is on monitoring of activities of daily living to support of industrial tasks such as aircraft maintenance.
A rich source of information, as yet unused in activity recognition, is the movement of the eyes. The movement patterns our eyes perform as we carry out specific activities have the potential to reveal much about the activities themselves - independently of what we are looking at. In this talk we report results from initial studies on activity recognition using a wearable electrooculography (EOG) system.