University of California, San Diego (UCSD) researchers are developing an algorithm that uses group pictures to determine what urban tribe a specific person belongs to, with up to 480percent accuracy.
The researchers say the algorithm could have a wide range of applications, from generating more relevant search results and ads, to allowing social networks to provide better recommendations and content.
"We are scratching the surface to figure out what the signals are," says UCSD professor Serge Belongie.
In developing the algorithm, the researchers decided to examine group pictures rather than pictures of individuals, a strategy they hope will make it easier to pick up social cues, such as clothing and hair styles, to determine people's tribes based on visuals featuring more than one person. The researchers designed the algorithm to analyze the picture as the sum of its parts and attributes, and to analyze the boxes for color, texture, and other factors. They then fed the algorithm images labeled for the urban tribes they represent, such as hipsters, surfers, bikers, and Goths. Finally, they fed the algorithm pictures free of labels.
The researchers say they are now working to improve the analysis of facial features and other attributes within the system.
From UCSD News (CA)
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found