If a human will see two photos with unknown persons he will recognize the same person right 97.53 percent of the time. Facebook's AI research group developed the software DeepFace which was right 97.25 percent on the same challenge, regardless of variations in lighting or whether the person in the picture is directly facing the camera. The software DeepFace will be presented on IEEE Conference on Computer Vision and Pattern Recognition in June 2014.
We closely approach human performance. (Y. Taigman, a member of Facebook's AI team)
The results are based on the power of a new approach to artificial intelligence known as "Deep Learning". It uses artificial neuronal networks to learn to recognize patterns in large amounts of data.
The right recognition of unknown faces is a great step in comparison to today commonly used approaches based on face recognition databases. This kind of technology will change video surveillance and automated photo analysis.
Photos may contain the same meta data like email communication or SMS. If person A and person B were more than once recognized in the same photo a connection between both persons can assumed in same way like an email contact between A and B.
