Monday, May 05, 2008
Computer Studies Building, Room 703
University of Massachusetts
Discriminative Probabilistic Models for Robust Scene Text Recognition
A first-grader can still read better than the best character
recognition software. Why? The reasons for this are varied, but one limitation of many approaches is a feed-forward, bottom-up, pipelined architecture that isolates the many tasks and information involved in reading. The result is a system that commits errors from which it cannot recover and has components that lack access to relevant information. In this talk, we'll look at what it might take to develop a "seeing eye computer" that can read to the blind.