Aude Oliva

Department of Brain and Cognitive Sciences

Massachusetts Institute of Technology

http://cvcl.mit.edu/Aude.htm

Friday, September 25

12:00-1:30 pm

A space-centered approach to scene recognition

The human brain has a remarkable ability to understand the meaning of

visual environments, novel places, and events with a single glance.

Whether walking down the street of an unfamiliar city, flipping quickly

through television channels, or watching the fast cuts in movie

trailers, we experience the phenomenon of rapid scene understanding

without exerting the slightest effort. In this talk, I will present

results from behavioral, computational modeling, and neuro-imaging

studies which show that global features related to the spatial layout,

volume, and content of real world scenes are the means by which the

human brain can quickly identify the meaning of complex, natural

environments without the need to segment the scene into parts and

without relying on object recognition mechanisms.