A research team at the Chinese University of Hong Kong, led by Professor Xiaoou Tang, announced 99.15% face recognition accuracy achieved in Labeled Faces in the Wild (LFW) database (a database of face photographs designed for studying the problem of unconstrained face recognition).
The technology developed by Xiaoou Chen’s team is called DeepID, which is more accurate than visual identification.
LFW is the most widely used face recognition benchmarks. Experimental results show that, if only the central region of the face is given, with the naked eye in the LFW person recognition rate is 97.52%.
The three face recognition algorithms developed by Xiaoou Chen’s team now occupies the top three LFW recognition accuracy rate, followed by Facebook’s Deepface.
His lab has been based on the latest technological breakthroughs to produce a complete set of facial image processing system SDK, including face detection, face alignment of key points, face recognition, expression recognition, gender recognition, age estimation and etc.
Xiaoou Tang plans to provide face recognition technology for free to Android, iOS and Windows Phone developers; with the help of this FreeFace-SDK, the developer can develop a variety of applications based on face recognition on the phone. In addition, he also wants to take advantage of user feedback to further improve the accuracy of the algorithm.