Facial recognition algorithms hit new accuracy highs in latest NIST test

Idemia correctly matched 99.88 percent of 12 million faces in the mugshot category of the latest 1:N test by the U.S. National Institute of Standards and Technology.

Other developers topping individual performance categories include CloudWalk (mugshot profile, visa border, visa kiosk, border at 10+ years, Sensetime (mugshot n=16 million, mugshot webcam), NEC (mugshot at 12+ years). Algorithms from those developers and Paravision make up the top 12 in the corresponding leaderboard update.

New algorithms were also submitted for the twenty-first update of the 1:N FRVT by returning developers Dermalog, Innovatrics, Rank One Computing, Sensetime and others. NIST evaluated more than 300 algorithms from over 100 developers in total.

The February 10 Face Recognition Vendor Test report shows incremental improvements by the most recent algorithms in enrollment performance, as well as accuracy.

NIST updated its digital identity guidelines with new requirements for biometrics last December.

More recently, DHS S&T published the results of its 2022 Biometric Technology Rally, which tests system accuracy in a simulated field test.

Article Topics

accuracy  |  algorithms  |  biometric testing  |  biometrics  |  Face Recognition Vendor Test (FRVT)  |  facial recognition  |  IDEMIA  |  NEC  |  NIST  |  Paravision