I came across a paper the other day by Serge Belongie's group at UCSD
which has a simple, yet clever, solution of how to incorporate "common
sense" information about what objects belong in a scene to improve
computer vision algorithms. Basically, if you have an idea of some of
the objects detected in a scene, you can use Google Sets to increase
the probability of detecting related objects. For example, it's easier
to determining if an object is a lemon or a tennis ball if you're also
detecting a tennis racket and person in the same image.

http://www.cs.ucsd.edu/~sjb/iccv2007a.pdf
http://www.sciencedaily.com/releases/2007/10/071017174328.htm
http://labs.google.com/sets

The research is still quite early, but could Google Sets also be
useful for more general AI tasks?

-- Neil

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=63889787-7348b4

Reply via email to