Eye Tracking : Opportunities for Parallelization

Eye Tracking : Opportunities for Parallelization

In this presentation I will provide some background on eye tracking, and discuss opportunities for supercomputing and cyberinfrastructure solutions within this exciting interdisciplinary domain. Eye tracking is a research area concerned with the measurement of eye movements to give us an idea of how humans deploy visual attention about the scene.
The technical approach to this measurement typically involves computer vision methods to image the pupil and limbus as the eye rotates. Eye tracking applications can in general be split along the real-time and off-line divide, by considering the former as interactive and the latter as diagnostic. Two ways in which the display can be affected in real-time is via gaze-contingent display or via gaze-based selection. Diagnostic evaluation of eye movement provides insight into human performance via examination of how some action was performed (e.g., visual inspection), and can also provide analysis of the stimulus under inspection. Web pages have received a lot of attention this way. We have made significant progress in each of these directions. We are now exploring ways in which these approaches can be improved by application of cyberinfrastructure and supercomputing methods.