Visual Representation and 3D Reconstruction of Dynamic Environments

GRAND Seminar March 21, Mon., 2011, ENGR 4201

Yaser Sheikh
http://www.cs.cmu.edu/~yaser/
Assistant Professor
Carnegie Mellon University

Host:

Jana Koseka

Abstract:

Visual reconstruction algorithms have largely focused on reconstructing static environments from multiple views. Real world scenes, however, are not static. Our social environments abound with moving objects: faces expressing and emoting, bodies moving and gesticulating, and vehicles locomoting. In this talk, I will discuss progress in developing a geometric framework for reconstructing dynamic environments inspired by the existing framework for multiple view geometry in static scenes. A central challenge is representing structure in the deformation of dynamic objects. I will present and motivate spatial and temporal models of objects, and motivate the use of spatiotemporal models for dynamic environments.

As the talk progresses, I will show applications of these models for augmenting hand animation with CG motion, photo-collection browsing, and motion capture editing.

If you would like to meet with the speaker contact kosecka@cs.gmu.edu.