The company Panasonic has created a security system that can recognize faces even if they are covered by glasses or even if they are in 10-year old photos
During Essen 2018, one of the most important world fairs on cybersecurity and scheduled in the German city from October 25, Panasonic will show its new facial recognition system developed in collaboration with Genetec Security Center.
Specifically, this is a platform that leverages deep learning algorithms and integrates sources from multiple devices (smartphones, tablets, surveillance cameras and so on) for facial recognition in large environments. From tests conducted over the past few years, the new facial recognition system developed by Panasonic is the most comprehensive in the world. The Japanese giant's tool can recognize faces from very complex angles, including faces tilted at an angle of up to 45 degrees to the left or right or 30 degrees up or down. And it has no particular problem detecting faces partially hidden by sunglasses or masks, either.
How Panasonic's innovative facial recognition works
Panasonic's new facial recognition system is designed for the security of private buildings and homes, and was also created with the collaboration of the National University of Singapore. The transmission of images captured by Panasonic's iPro Extreme video surveillance system that includes this new face recognition is in real time, and each individual server can receive and analyze incoming video from 20 different cameras at the same time through the use of artificial intelligence. According to Panasonic's statement, the new face recognition is able to identify a person not only live but also by analyzing a photograph taken up to 10 years earlier.
Panasonic's security platform is already being used in Japan at Tokyo's Haneda Airport where, after three years of field testing, it has enabled it to replace fingerprint recognition operations at all boarding gates. Finally, according to recent Panasonic releases, the newly developed face recognition system has a 90% success rate in identifying people from video and should have no problems with skin color differences typical of other similar systems.