ROC tops NIST fingerprint identification benchmark

New results from NIST’s Friction Ridge Image and Features Technology Evaluation One-to-Many (FRIF TE E1N) place ROC at the top of the benchmark’s Class B slap fingerprint accuracy rankings.

The results come as biometric vendors increasingly use NIST benchmark evaluations to demonstrate performance to government agencies and enterprise buyers evaluating automated biometric identification systems (ABIS).

ROC’s algorithms returned the lowest error rate across a customary cohort of Class B identification error rate metrics, while the company also scored top accuracy for False Negative Identification (FNIR) at Rank-12 for the same Class B evaluation criteria.

“In just four months, ROC went from a strong first submission in FRIF E1N to the best in the world across substantial portions of the evaluation,” says Roc’s Principal Scientist Josh Engelsma. “This is only our second submission, and as a highly focused American biometrics company, ROC still has a lot of gas left in the tank for what comes next.”

The results could bolster Roc’s push into government and security markets following its February IPO. NIST’s FRIF TE E1N is an important reference for governments and enterprises choosing large-scale Automated Biometric Identification Systems (ABIS), the Colorado firm notes.

“For years, the United States has maintained a dangerous overreliance on foreign AI for our most critical identity and biometric screening systems,” says the company’s CEO, B. Scott Swann. “ROC’s historic performance in the NIST fingerprint evaluation proves America now has a world-class domestic alternative.”

The open-set identification evaluation tests algorithms that automatically extract and use features from all types of exemplar friction ridge images, including rolled fingerprints, palm prints and slaps. Those features are used to search for similar candidates in databases of millions of subjects.

Class B is a particularly complex test that evaluates slap fingerprint accuracy using simultaneous multi-finger impressions from 4-4-2 captures, including left slap, right slap and both thumbs.

The company has published the full test results on its blog.

Other biometrics firms, including Neurotechnology, Tech5 and Dermalog, have also recently published results from their NIST testing.

Earlier in May, Neurotechnology shared that it has achieved strong results in the FRIF TE E1N testing, ranking in first place in accuracy in two out of three categories and achieving performance that is more than 200 times faster in Class C.

Tech5 reported a 0.0-second search time on a gallery of five million fingerprint records, demonstrating matching speeds in the millisecond range.

Dermalog also achieved notable results in Class B, becoming one of only two participants to reach zero false-negative identifications (FNIR) for matching identification flats at a false-positive identification rate (FPIR) ≤ 0.001.

The latest results also highlight the growing competition among fingerprint identification providers seeking to differentiate themselves through independent NIST testing. These recent submissions from ROC, Neurotechnology, Tech5 and Dermalog underscore how benchmark evaluations are increasingly serving as a key reference point for organizations procuring large-scale biometric identification systems.

Article Topics

biometric identification  |  biometrics  |  fingerprint biometrics  |  fingerprint recognition  |  Friction Ridge Image and Features (FRIF)  |  NIST  |  ROC