Event Location: NSH 1507
Bio: Todd Zickler received the B.Eng. degree in honours electrical engineering from McGill University in 1996 and the Ph.D. degree in electrical engineering from Yale University in 2004. He subsequently joined the Harvard School of Engineering and Applied Sciences, where he is currently an associate professor. Todd’s interests span computer vision, image processing, computer graphics, and human perception; and much of his work is devoted to developing efficient representations for appearance and exploiting them for visual tasks. He received an NSF career award in 2006 and was named an Alfred P. Sloan research fellow in 2008. His work is funded by the NSF, ARO and ONR. More information can be found on his web-site: http://www.eecs.harvard.edu/~zickler.

Abstract: A vision system is tasked with inferring the observable properties of a scene—shape, materials, and so on—from one or more of its images. The task is made hard by the fact that the mapping from scene properties to images is many-to-one: For any given image, there are infinite scenes to explain it.

A viable approach for dealing with this ambiguity is designing systems that combine prior visual experience with loose, redundant constraints induced by texture, shading, and various other aspects of optical stimulation. The basic idea is that each cue reduces the set of interpretations in some way, and by combining them, systems will be better equipped to sift through the infinite set of possibilities and arrive at a reasonable result.

Important to the performance of such visual inference engines is an understanding of the many different ways in which shape and materials are encoded in image data. In this talk I will focus on two of them, each of which exists in the presence of complex “real-world” lighting. For each case, I will summarize our progress toward characterizing the constraints induced on a scene, and our progress in creating algorithms for inference.