Communications of the ACM
What Makes Paris Look Like Paris? CM Software ncovers Stylistic Core
Credit: Carnegie Mellon University/Google Street View
Carnegie Mellon University (CMU) researchers have developed visual data-mining software that can automatically detect the subtle features that make cities unique, such as street signs, street lamps, and balcony railings.
The software analyzed more than 250 million visual elements taken from 40,000 Google Street View images of Paris, London, New York City, Barcelona, and eight other cities around the world to find those features that could be used to distinguish one city from the rest. The software found sets of geo-informative visual elements unique to each city, such as cast-iron balconies in Paris and fire escapes in New York. The researchers presented their findings at the SIGGRAPH 2012 conference.
"Our data-mining technique was able to go through millions of image patches automatically--something that no human would be patient enough to do," says CMU professor Alexei Efros. The software had more trouble identifying geo-informative elements in U.S. cities, which the researchers say is due to the relative lack of stylistic coherence in American cities.
"In the long run, we wish to automatically build a digital visual atlas of not only architectural but also natural geo-informative features for the entire planet," Efros says.
From Carnegie Mellon News (PA)
View Full Article
Abstracts Copyright © 2012 Information Inc., Bethesda, Maryland, USA
No entries found