Paravision, a provider of facial recognition software, has announced that it was ranked as the most accurate facial recognition algorithm in the January 2022 1:N facial recognition vendor test (FRVT) conducted by the US National Institute of Standards and Technology (NIST).

1:N test measures the facial recognition algorithm’s ability to match a photo from a gallery of images, which is evaluated by NIST for accuracy in facial recognition for borders/visa purposes, comparing a large dataset of images with challenging environments, such as different posing angles, low resolution, an low contrast.

Paravision’s algorithm has reduced its error rate by more than 95% over the past three years, recording a reduction of 70% within just one year. The error rate is recorded at 0.22% for false-negative identification and 0.3% for false positives. 

Analyst QuickTake: 1:N category is the most used facial verification technique for border control and law enforcement, as it involves matching photos with a large gallery of images. Being ranked as the most accurate algorithm would help Paravision to further penetrate the surveillance and law enforcement markets, competing with peers like Clearview AI.