IF TWO FACES of the same person are presented side by side, there’s a 97.53 per cent chance you will get it right. A machine, however, might have a more difficult time achieving the same results until now.
Now Facebook has said it’s developed a technology that is 97.25 per cent accurate in recognising faces. The project, named DeepFace, maps out 3D facial features, before making a flat model that’s filtered by colour to characterise specific facial elements.
The company has also said it’s tapped into a pool of 4.4 million labeled faces from 4,030 different people on its network to help the system learn.
According to MIT Technology Review, the research project isn’t ending up on Facebook, but will be released ahead of presenting it at the IEEE Conference on Computer Vision and Pattern Recognition this June, to get feedback from the research community.
This project isn’t a surprise by any means since Facebook has been experimenting with facial recognition for a while. It introduced a form of facial recognition back in 2010 which identified faces but that ended up in a audit by the Irish Data Protection Commissioner.
Also, the company purchased Israeli startup Face.com in June 2012, a startup that developed facial recognition technology for mobile devices, for $60 million.
To embed this post, copy the code below on your site
have your say