CPS Seminar Speaker Jitendra Malik SEA 4.244

Contact Name: 
Jenna Whitney
Date: 
Nov 27, 2006 12:00pm - 1:00pm

Speaker/Affiliation: Jitendra Malik Ph.D. Profes

sor Electrical Engineering and Computer Sciences University of California
Berkeley

When/Location: 11/27/06 12:00 PM SEA 4.244

Title
of Talk: Probabilistic models of perceptual organization

Reception

with Refreshments at 11:30 AM

Abstract: Visual grouping and figure-g

round discrimination were first studied by the Gestalt school of visual per

ception nearly a century ago. By the use of cleverly constructed examples

they were able to demonstrate the role of factors such as proximity simila

rity curvilinear continuity and common fate in visual grouping and factors
such as convexity size and symmetry in figure-ground discrimination. How

ever this left open (at least) three major problems
(1) there wasn''t a
precise operationalization of these factors for general images
(2) the
interaction of these cues was ill understood
(3) and there was no justi

fication for why these factors might be helpful to an observer interacting

with the visual world.

Over the last few years we have been pursuin

g these problems in the following paradigm:
(1) We start with a set of n

atural images and use human observers to mark the perceptual groups and ass

ign figure-ground labels to the various boundary contours.
(2) We constr

uct computational models of various grouping and figure-ground factors.

(3) We calibrate and optimally combine the grouping and figure-ground facto

rs by using the principle that vision evolved to be adaptive to the statist

ics of objects in the natural world.

In my talk I will report on two
recent results in this paradigm. One is on understanding the power of the

figure-ground cues specifically size lower-region and convexity. We compa

red the predictions of such a model with pyschophysics and found a pleasing
agreement. The second is an attempt at a unified probabilistic framework f

or mid-level vision using conditional random fields defined on constrained

Delaunay triangulations of image edges.

This talk draws on joint wor

k with Charless Fowlkes David Martin and Xiaofeng Ren; various papers can
be found on the web site http://www.eecs.berkeley.edu/Research/Projects/CS

/vision/grouping