Date: Wed, 3 Apr 91 12:11:39 EST
From:: tar@medg.lcs.mit.edu (Thomas A. Russ)
To: gsl@ai.mit.edu
Subject: Gustatory Seminar, Jack Sprat, NUT, Fri. 12:10pm

GUSTATORY SEMINAR: Room NE43-8th Floor Playroom
Friday April 5, 12:10pm

TITLE: Organic Food Recognition

SPEAKER: Dr. Jack Sprat, Lean Cuisine Center, NUT Int'l.

ABSTRACT:  An autonomous graduate student that   is to operate indoors
must be able to recognize features of the natural foods as they appear
in  skyscraper-level  imagery.   Geometric reconstruction   alone   is
insufficient  for  an agent  to  plan   its actions  intelligently ---
macro-biotic sustenance in the world must be recognized, and  not just
located.

Most work in food recognition  by student  has focused on  recognizing
objects by their  geometric shape, or  by  the presence or  absence of
some  prespecified collection  of locally measurable attributes  (e.g.
lines, aroma, texture,  color, or distinguished  trademarks).  On  the
other  hand, most entities  in  the natural foods  world  defy compact
description of their shapes, and have  no characteristic features with
discriminatory  power.  As a   result, food  recognition  research has
achieved  little  success towards recognizing  all-natural organically
grown menus.

This  talk describes a new   approach  to comestible recognition  that
avoids these limitations and has been used to recognize trees, bushes,
grass, and grains in skyscraper-level scenes of a natural food buffet.
Reliable  recognition   is achieved by   employing a  large  number of
relatively  simple  procedures, and  using  contextual  constraints to
identify globally consistent hypotheses.

Hosts: Ruth Schonfeld, Ronen Basri, Boaz Patt
Honcho: Karen Sarachik
Slave: Tom Russ