Event Location: NSH 1507
Bio: Pedro Felzenszwalb is an Assistant Professor at the Department of Computer
Science at the University of Chicago. He received his PhD degree from MIT
in 2003. His research focuses on computer vision, algorithms and
artificial intelligence.

Abstract: I will describe a new representation for two-dimensional objects that
captures shape information at multiple levels of resolution. The
representation is based on a hierarchical description of an object’s
boundary, and can be used in an elastic matching framework, both for
comparing pairs of objects and for detecting objects in cluttered images.
I will show how these problems can be solved efficiently using a dynamic
programming procedure similar to CKY parsing.

In contrast to classical elastic models, our representation explicitly
captures global shape information. This leads to richer geometric models
and more accurate recognition results.

I will also discuss how the hierarchical representation can be used in a
part-based modeling scheme, where object models can share common parts.
Moreover, the parts can share subparts and objects can have variable
structure. This leads to the notion of a single context-free grammar that
compactly represents a large set of objects.