VASC Seminar Announcement
            =========================

Date: Friday, 11/19/99
Time: 10:30-11:30
Place: Smith Hall 2nd Floor Common Area

Speaker:  Bill Freeman
          Mitsubishi Electric Research Labs
          http://www.merl.com/people/freeman/

Title:  Learning low-level vision

Abstract:

We describe a learning-based method for low-level vision
problems--estimating scenes from images.  We generate a synthetic
world of scenes and their corresponding rendered images, modelling
their relationships with a Markov network.  Despite the presence of
loops in the Markov network, we find that the belief propagation
yields good estimates for the MAP scene estimate, and discuss why
(work with Yair Weiss).  We call this approach VISTA--Vision by
Image/Scene TrAining.

We apply VISTA to the ``super-resolution'' problem (estimating high
frequency details from a low-resolution image), showing good results.
To illustrate the potential breadth of the technique, we also apply it
in two other problem domains.  We learn to distinguish shading from
reflectance variations in a single image (of a constrained class of
images).  For the motion estimation problem, we show figure/ground
discrimination, solution of the aperture problem, and filling-in
arising from application of the same probabilistic machinery.