Scientists inculcate common sense to computers

Common sense is not common. It is true. The American scientists are now using Google Labs widget to inculcate common sense to a computer so that it will enable automated labeling system to identify objects in photographs.

coumuters get common sence
coumuters get common sence

The basic function of an automated object identifier, as we know is to label an object in a photo. The object may be a person, a tennis court, a tennis racket or anything else. The scientists from University of California, San Diego and Los Angels have now developed a computer-based automated labeling system to re-label an object. It means that you can re-label the lemon as a tennis ball and vice-versa. The new system is also called post-processing context.

They have used Google Labs tool, Google Sets to provide external contextual information to automated object identifiers.

Serge Belongie, a computer scientist from UC San Diego commented,

In some ways, Google Sets is a proxy for common sense. In our paper, we showed that you can use this common sense to provide contextual information that improves the accuracy of automated image labeling systems.

He has explained three systematic steps required for image labeling system. The first step is image segmentation that enables the automated system to split the image into different regions. The second step is ranking the list of probable labels for the image regions.

Lastly, the system integrates a quantity of context by processing all the different possible combinations of labels within the image and maximizing the contextual agreement among the labeled objects within each picture.

Via: Science Blog

You might also like
© 2012, Instamedia. Some rights reserved. Powered by Instapress and Instacheckin.