Acknowledging the Third Dimension
ABSTRACT:Successful vision algorithms must go beyond local analysis within the 2D image plane, into a global analysis of objects and surfaces within the context of the 3D scene. But how can we do all of this if we only have only a single image? I will describe our first steps, in which we provide a rough layout of the major surfaces and perspective of the scene. Such information can then be used to improve object recognition, navigation, and other real world tasks.
Derek is a graduating (this summer) Ph.D. student at at the Robotics Institute in Carnegie Mellon University,
co-advised by Alyosha Efros and Martial Hebert . Derek received B.S.'s in electrical and computer engineering
from SUNY at Buffalo in 2002 and a M.S. in robotics from CMU in 2004. His research interests are in complete
visual scene understanding and the related computer vision, machine learning, and graphics techniques and applications.
Last modified: Fri Aug 8 11:38:17 CST 2006