Iris ID debuts in NIST FRTE 1:1, ROC impresses with updated algorithm

The latest edition of NIST’s ongoing report series evaluating biometric face verification accuracy includes the debut of algorithms from Iris ID and Vietnam-based FPT Telecom JSC, along with new submissions from half a dozen returning developers.

The top performers in Monday’s update of the FRTE 1:1 Verification report are largely unchanged from the previous report in January, with the exception of a strong performance from ROC’s new algorithm.

Other returning developers sending in new technology include India-based Biometrik Systems, Vietnamese infrastructure software provider GTEL OTS, id3, big data analytics and computer vision provider NextMeta Labs and Nominder.

The ROC_020 algorithm had the fourth-lowest false non-match rate (FNMR) in the MUGSHOT-MUGSHOT category, and fifth-lowest in the BORDER-BORDER category. It also placed in the top ten in the MUGSHOT ΔT≥12 YRS, VISA-BORDER and BORDER-KIOSK category. Its lowest showing was 14th in the VISA Yaw≥45- BORDER° category.

Iris ID’s results placed it just ahead of the middle of the pack among the more than 350 facial recognition algorithms submitted to the FRTE 1:1, topping out at 112th in the BORDER-BORDER category with an FNMR of 0.0108 at the default false match rate (FMR) of 0.000001.

FPT Telecom’s accuracy results were more modest.

QazSmartVision, STCon, Viante, Paravision, Cloudwalk and Innovatrics remain among the most accurate submissions tested by NIST.

Article Topics

biometric testing  |  biometrics  |  Face Recognition Technology Evaluation (FRTE)  |  facial recognition  |  Iris ID  |  NIST  |  ROC