Talk by Paulo E. Santos:
Cognitive Vision in Movement and Action
Date : Friday, January 26, 2007
Time : 3-4pm
Room : PT 266
Abstract:
In this talk I present two approaches for knowledge assimilation from
vision data: the first includes a novel logic-based formalism for
representing knowledge about objects in space and their movements
(including the observer). In this framework space is represented using
functions that extract attributes of depth, size and distance from
snapshots of the world. The second part of the talk presents a cognitive
vision system capable of autonomously learning protocols from perceptual
observations of dynamic scenes. These approaches are motivated by the aim
of creating a synthetic agent that can interpret scenes containing
interactions between unknown objects and agents, and learn models of these
that are sufficient to allow the agent to act in accordance with the
implicit protocols present in the scenes.
Paulo E. Santos, Ph.D.
Dept. of Electrical and Electronics Engineering
Centro Universitario da Fei
Sao Paulo, Brazil
http://www.fei.edu.br/~psantos/
