NIST facial recognition evaluations showcase accuracy gains, new developers – Biometric Update

Asian biometrics developers continue to join the U.S. National Institute of Standards and Technology’s (NIST’s) facial recognition tests, with three new companies contributing algorithms to the latest evaluations.

Both the 1:1 Verification and 1:N Identification tracks of the Face Recognition Technology Evaluation (FRTE) were updated with new reports on July 3. The reports show continued improvement in match rates overall, and the top performers are mostly the same as in other recent editions of the reports. A relative newcomer in the top of the 1:1 Verification leaderboard is Dubai developer Viante.AI, which placed in the top 5 in several categories with an algorithm submitted in June.

The new 1:N report includes a submission from Sansap Technology, an India-based developer marking its entry into NIST evaluations, as well as new algorithms from Alchera, Clearview AI, CyberLink, Dermalog, Nominder, Omnigarde, and ROC.

Innovatrics is pleased with performance gains from its algorithm submitted in May. In particular, the company highlights that it placed 12th in the Visa-Border and 11th and 13th in the Mugshot-Mugshot categories for databases with 12 million and 16 million subjects, respectively.

Paravision also jumped back into the top ten in a number of categories with an algorithm submitted in June.

Several of the same returning developers have submitted new algorithms for evaluation in the 1:1 report, as well as Facephi.

New face biometrics verification algorithms are included from Taiwan Computer Vision (a subsidiary of Japan Computer Vision) and Taiwan-based Telexper International.

Omnigarde claims that the 1:1 evaluation shows it third among U.S.-based developers, while placing it 11th in matchng accuracy for the Mugshot-Mugshot dataset. The Omnigarde-004 algorithm improved performance in average error rate by 25 percent over its predecessor.

Ominigarde also placed just ahead of and behind Innovatrics in the 1:N Mugshot-Mugshot categories.

Article Topics

accuracy  |  biometric testing  |  biometrics  |  Face Recognition Technology Evaluation (FRTE)  |  facial recognition  |  NIST

This post was originally published on 3rd party site mentioned in the title of this site