Sberbank patents new method to augment facial recognition for crowds
PJCS Sberbank has been granted a new patent for a system that applies augmented reality and 3D modeling to improve biometric facial recognition performance in identifying individuals in crowds, spotted by Communal News.
The method involves faces being identified in a region of space, and then various images of each face being analyzed for the amount and quality of data to chose the best image for comparison, according to the patent, as translated by Google.
The system supports the use of AR glasses, according to the report, and can trigger an alert, for instance if a blacklisted individual is identified. The patent describes the use of a web interface to edit descriptions and add people to the database or edit the blacklist.
Sberbank is the owner of the Speech Technology Center, which provides facial recognition among a range of biometric technologies, after acquiring a controlling stake in the company last year. The technology could have immediate practical benefits for the STC, which recently deployed its facial recognition to Moscow’s VTB Arena to identify individuals banned from attending events in the 26,000-seat facility.
Last year, Chinese researchers chose a different approach to the same problem, developing a 500 megapixel camera to capture high-resolution images capable of supporting facial biometrics matching in the cloud.