Suprema has shared a blog post that explains how infrared technology can be used to build more robust facial recognition systems. The company currently uses Near Infrared (NIR) tech in its FaceStation 2 and F2 facial recognition terminals, and eventually hopes to reduce the computational footprint enough to run on smartphones and other edge devices.
As it stands, most smartphones feature simpler 2D facial recognition system that rely only on standard RGB camera technology. Such algorithms are small enough to run on most consumer devices, but they are not as secure and can be bested with photos and other common spoofing techniques. They also struggle when asked to recognize new features on a person’s face, such as a mask or a new pair of glasses.
3D mapping systems like the one found in the iPhone 12 are more effective against spoofing. However, they are still quite expensive, and are only found in high-end devices as a result. Suprema’s system, on the other hand, uses NIR tech to supplement a more conventional color scan. In that regard, it captures multiple images of a person’s face, and uses AI to bridge the gap and generate a single template of the user. More specifically, AI maps the NIR data onto the color facial image, turning it into a unique feature that prevents spoofing and gives the system more confidence when performing a facial recognition match.
Suprema’s solution offers on-device facial recognition, and can function with any camera with 720p resolution. That should make it a reasonably popular option if Suprema is able to reduce the power consumption enough to run on battery-powered IoT devices.
In the meantime, organizations are encouraged to use other biometric modalities (such as fingerprint recognition) to safeguard more sensitive assets. Suprema’s BioSign 4.0 fingerprint algorithm recently appeared in Samsung’s Galaxy S21 smartphone.
Source: Android Authority
July 20, 2021 – by Eric Weiss