Lecture 19 - October 30, 1997

This lecture continues discussion of connectionist vision models, started in Lecture 16 and includes some material on Markov Random Fields.

The best reference for Cooper's work on Markov Random Fields and the recognition of occluded shapes is:

Cooper, P.A. "Parallel Structure Recognition with Uncertainty" Proc. ICCV, Osaka, Japan , December 1990

The best reference for Goddard's work on Dynamic Connectionist Models and the recognition of human gaits is:

Goddard, N. H. "Human Activity Recognition" in Motion-Based Recognition, Shah, M. and Jain, R. (Eds.), Kluwer Academic Publishers, 1997.

There will be a related lecture by John Hummel in the Cognitive Science Colloquium on November 7 at 11am in 2515 Tolman Hall. A good reference for this work is:

Hummel, J. E., & Biederman, I. (1992). Dynamic binding in a neural network for shape recognition. Psychological Review, 99, 480-517.