One of the most famous face recognition systems is Apple Face ID. It uses special illumination from infrared points, which are allowed to map the depths of the user’s face.
The program is not affected by bright makeup, bushy beard and even sunglasses, but it cannot cope with medical masks. The problem can be solved by using the little-known Express Transit feature, which allows you to use Apple Pay on public transport without authentication.
Unfortunately, not all countries and cities support Express Transit. In May, Apple released an update that quickly detected the presence of a mask and immediately prompted for a password.
According to Andrew Bud, CEO of facial recognition company iProov, Express Transit may be the best solution in the near future.
Tech companies are adapting to this reality. However, experts believe that users themselves will have to change first.
The main problem is that it is difficult for the system to distinguish between a real masked face and a simulated face. And verifying the reality of the face is at the heart of the security of face verification.
“Modern programs are predominantly focused on the eye area,” explains Bud. “The previous solution – measuring the geometry of the face as a whole – was outdated five years ago.”THE MASKS BY THEMSELVES DO NOT HARD FACE RECOGNITION
The issue of vitality has also jeopardized competitive developments. A month after Samsung released the Galaxy S8, independent researchers posted a video in which they unlocked the phone using an infrared photograph of the iris and a contact lens. “The most expensive part of hacking biometrics was buying a Galaxy S8,” they said.
There is also a positive side to the difficulty in recognizing masked people. The problem faced by authentication services is to be solved by facial recognition software. In July 2020, the US National Institute of Standards and Technology published a report on the impact of having masks on identification accuracy. The results of the study showed that the presence of masks reduced the accuracy of face identification algorithms by 5-50%.