Communications of the ACM
Smart Computer Learns From Video
The ETH Zurich researchers' program recognizes the spatial and temporal patterns of sequences of activities in road traffic.
Credit: Daniel Kttel / ETH Zurich
Swiss Federal Institute of Technology Zurich (ETH Zurich) researchers have developed a learning program that can analyze temporal and spatial patterns of moving objects. The software can analyze street scenes from video, map the patterns that characterize the various road users, and establish rules governing the traffic flow. Once the program has "learned" the standard patterns, it can interpret video input in real time. "The hardest part of it was processing the theory behind it," says ETH Zurich researcher Daniel Kuttel.
Automatic interpretation of dynamic camera images could help with traffic management control centers that monitor traffic flow, says ETH Zurich professor Vittorio Ferrari.
As an extension of the research, the ETH Zurich team will attempt to train the program to recognize visual concepts. For example, the program would be able to search the Internet for specific images and automatically find the correct image, including those that are not labeled correctly.
From PhysOrg.com
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA 
No entries found