Facebook's algorithm delivers an accuracy rate of 83%
Yann LeCun, head of the Artificial Intelligence department at Facebook, has presented an experimental algorithm at the Computer Vision and Pattern Recognition conference in Boston that can recognize people even if their faces are covered up.
"There are a lot of cues we use. People have characteristic aspects, even if you look at them from the back," said LeCun for science news website New Scientist.
Testing the system with over 40,000 photos pulled from Flickr, Facebook reported an accuracy rate of 83%, which is quite impressive for the task it was programmed to carry out.
The algorithm is, of course, still not yet live on the site, but if employed, it can seriously improve usage of the tagging feature, which even if quite useful, still relies on human users to do the tagging.
Full article. http://news.softpedia.com/news/expe...l&utm_source=facebook.com&utm_campaign=buffer
Yann LeCun, head of the Artificial Intelligence department at Facebook, has presented an experimental algorithm at the Computer Vision and Pattern Recognition conference in Boston that can recognize people even if their faces are covered up.
"There are a lot of cues we use. People have characteristic aspects, even if you look at them from the back," said LeCun for science news website New Scientist.
Testing the system with over 40,000 photos pulled from Flickr, Facebook reported an accuracy rate of 83%, which is quite impressive for the task it was programmed to carry out.
The algorithm is, of course, still not yet live on the site, but if employed, it can seriously improve usage of the tagging feature, which even if quite useful, still relies on human users to do the tagging.
Full article. http://news.softpedia.com/news/expe...l&utm_source=facebook.com&utm_campaign=buffer
