Invision: a System Using Vision Patterns to Understand User Attention Jul Iraz001 Barker L Braries Invision: a System Using Vision Patterns to Understand User Attention

This research demonstrates a new approach that uses patterns of eye motion for object selection in an eye tracking interface. The analysis of eye motion on the pattern level can deliver three values to an eye tracking interface: selection speed, reliability, and a more complete understanding of user attention. Current eye tracking interfaces use fixation as a means of target acquisition and/or selection. There are several problems with this approach concerning issues of selection speed, system reliability and the understanding of user attention. This research builds a system, called InVision, to demonstrate how the analysis of eye fixation at the pattern level can help provide solutions to these problems. First, selection speed is quick through the use of pattern identification as a means of selection. Second, pattern correlation can add reliability to an eye tracking interface. Finally, the ability to understand the context of a user's eye motion is provided through pattern interpretation. Thesis Supervisor: Ted Selker Title: Associate Professor, MIT Media Laboratory