ROC sweeps Class B in NIST’s latent biometric accuracy assessment

ROC is making the case for U.S. domestic fingerprint biometrics at scale to be powered by algorithms developed in America, armed with improved results in its latest evaluation by NIST. The company emphasizes the multiple categories in which it scored the highest accuracy in the Evaluation of Latent Fingerprint Technologies ( ELFT ), along with the fast speed of its latent print processing. The latest ELFT results place ROC first in all 28 Class B Metrics. Class B is the set of fingerprint 4-4-2 capture scenarios, those involving “slap” scanners for simultaneous multi-finger biometrics collection. The company’s strong ELFT results follow a strong showing in NIST’s FRIF TE E1N , particularly in the Class B categories. ROC’s algorithm also performed 64-times faster in the ELFT evaluation than the mean processing time of the top five vendors evaluated across datasets from the FBI and DoD, at 102 seconds. Principal Scientist Josh Engelsma notes in a post analyzing the company’s results that “large-scale U.S. biometric identity systems have relied heavily on foreign fingerprint providers.” The latest results show ROC’s readiness to replace those foreign companies as an automated biometric identification system (ABIS) supplier for American government and law enforcement agencies operating high-throughput systems. To that end, the company’s ROC Watch was also granted Developmental Testing and Evaluation (DT&E) Designation by DHS just days ago. “Latent fingerprint recognition is one of the most technically challenging evaluations in biometrics, due to distorted or partial evidence utilized in analysis against sizable identity databases,” ROC CEO B. Scott Swan. ROC also tied for the best accuracy among biometrics providers from the Western hemisphere in Rank-1 results for the FBI Laboratory Dataset with its latest algorithm , at 0.0612. “What stands out in this ELFT submission is ROC’s optimization of both accuracy and throughput,” says ROC Chief Scientist and Co-founder Dr. Brendan Klare. “The low error rates across multiple benchmark conditions, combined with the fastest search speed, shows continued leadership and progress on the two outcomes that materially affect large-scale latent search: candidate quality and system throughput. These gains are especially meaningful in large-scale deployments, where performance, compute, and total cost of ownership must be considered together.”

ROC is making the case for U.S. domestic fingerprint biometrics at scale to be powered by algorithms developed in America, armed with improved results in its latest evaluation by NIST.The company emphasizes the multiple categories in which it scored the highest accuracy in the Evaluation of Latent Fingerprint Technologies (ELFT), along with the fast speed of its latent print processing. The latest ELFT results place ROC first in all 28 Class B Metrics. Class B is the set of fingerprint 4-4-2 capture scenarios, those involving “slap” scanners for simultaneous multi-finger biometrics collection.

The company’s strong ELFT results follow a strong showing in NIST’s FRIF TE E1N, particularly in the Class B categories. ROC’s algorithm also performed 64-times faster in the ELFT evaluation than the mean processing time of the top five vendors evaluated across datasets from the FBI and DoD, at 102 seconds. Principal Scientist Josh Engelsma notes in a post analyzing the company’s results that “large-scale U.S. biometric identity systems have relied heavily on foreign fingerprint providers.

”The latest results show ROC’s readiness to replace those foreign companies as an automated biometric identification system (ABIS) supplier for American government and law enforcement agencies operating high-throughput systems. To that end, the company’s ROC Watch was also granted Developmental Testing and Evaluation (DT&E) Designation by DHS just days ago.

“Latent fingerprint recognition is one of the most technically challenging evaluations in biometrics, due to distorted or partial evidence utilized in analysis against sizable identity databases,” ROC CEO B. Scott Swan. ROC also tied for the best accuracy among biometrics providers from the Western hemisphere in Rank-1 results for the FBI Laboratory Dataset with its latest algorithm, at 0.0612.“What stands out in this ELFT submission is ROC’s optimization of both accuracy and throughput,” says ROC Chief Scientist and Co-founder Dr. Brendan Klare. “The low error rates across multiple benchmark conditions, combined with the fastest search speed, shows continued leadership and progress on the two outcomes that materially affect large-scale latent search: candidate quality and system throughput.

These gains are especially meaningful in large-scale deployments, where performance, compute, and total cost of ownership must be considered together.”