Microsoft is working on new face-recognition technology via its OneAlbum project being developed in Israel, which allows users to find photos across various social networks.
The OneAlbum project is under development by Microsoft Israel Innovation Labs. According to the project description, OneAlbum’s novel, unsupervised face-recognition and event-matching technology analyzes photos in a person’s album to find automatically, with no tagging required, the faces of people he or she cares about, based on the frequency of their appearance in the album.
Starting from technology developed by Microsoft Research Asia and Microsoft’s Live Labs, the Israel Innovation Labs built the new face-recognition technology.
OneAlbum crawls the photo albums of the user’s friends and social networks to find relevant photos. The algorithm has been tested on “real large-scale albums,” the site says, including those with “tens of thousands of photos” and has achieved accuracy rates of up to 90 percent.
“This,” says Eyal Krupka, principal research program manager for the Israel Innovation Labs, “includes photos of myself, my wife, and my kids, from the events I participate in, the places I like and more. OneAlbum also goes further, by organizing ‘my album’ – the photos from the same event are presented side-by-side, no matter where they come from.”