Non-rigid structure from locally-rigid motion

--------------------------------------------------------------------
                ***Joint Seminar***
--------------------------------------------------------------------
The Hong Kong University of Science & Technology

Department of Computer Science and Engineering
Center of Image, Vision and Graphics
--------------------------------------------------------------------

Speaker:	Prof. Kyros KUTULAKOS
		Computer Science, University of Toronto

Title:		"Non-rigid structure from locally-rigid motion"

Date: 		Monday, 12 April 2010

Time: 		2:00pm - 3:00pm

Venue: 		Room 3402 (via lifts 17/18)
		The Hong Kong University of Science & Technology


Abstract:

Non-rigidity is pervasive in the world around us: a person's body movements
and facial expressions, the deformations of cloth, and the collective motion
of a group (e.g., people, cars, plants, etc) can all be described as
non-rigid motions in 3D. Unfortunately, building non-rigid 3D models from a
video sequence has proved very hard, and still remains one of the few open
problems in visual reconstruction.

In this talk, I will present a new approach to the "non-rigid structure from
motion" problem that promises to significantly expand the non-rigid scenes
reconstructible from video in 3D. The idea is to first solve many local
3-point, N-view *rigid* reconstruction problems independently, providing a
"soup" of independently-moving and nearly-rigid 3D triangles. Triangles in
this soup are then combined into deforming bodies in a bottom-up fashion. I
will show results on a variety of challenging scenes, including deforming
cloth, tearing paper, faces, and multiple independently-deforming surfaces.

This is joint work with Jonathan Taylor and Allan Jepson at the University
of Toronto.


******************
Biography:

Kyros Kutulakos is a Professor of Computer Science at the University of
Toronto. He received a BA degree from the University of Crete and a PhD from
the University of Wisconsin, Madison, both in Computer Science. Following
his dissertation work, he held faculty appointments at the University of
Rochester and the University of Toronto, and a visiting scholar appointment
at Microsoft Research Asia.

Prof. Kutulakos is a recipient of an Alfred P. Sloan Research Fellowship, an
Ontario Premier's Research Excellence Award, and four best paper prizes (a
Best Paper Honorable Mention at the 2006 European Conference on Computer
Vision; a David Marr Prize Honorable Mention in 2005; a David Marr Prize in
1999; and an Outstanding Paper Award at the Computer Vision and Pattern
Recognition Conference in 1994). He was Program co-Chair of the
International Conference on Computational Photography in 2010 and of the
Computer Vision and Pattern Recognition Conference in 2003, and will be
Program co-Chair of the International Conference on Computer Vision in 2013.