ROC Earns #1 Global Ranking in NIST ELFT Latent Fingerprint Benchmark, Strengthening Its Position in Mission-Critical Identity Security
ROC delivers top Rank-1 accuracy and search speed on the DoD-provided dataset, supporting high-confidence searches at scale
Next-gen efficiency reduces infrastructure demands and total cost of ownership for real-world deployments by law enforcement, intelligence, and ABIS operators
DENVER, CO, March 02, 2026 (GLOBE NEWSWIRE) -- Rank One Computing Corporation d/b/a ROC, (NASDAQ:ROC) ("ROC" or the "Company"), a U.S. leader in multimodal Vision AI, building sovereign biometric, video analytics, and mission intelligence solutions into a unified platform, announces its latent fingerprint algorithm has been ranked #1 in performance with the lowest False Negative Identification Rate (FNIR) at Rank-11 on the U.S. Department of Defense (DoD)-provided dataset, the largest dataset in the latest National Institute of Standards and Technology (NIST) Evaluation of Latent Fingerprint Technologies (ELFT). These results reinforce ROC's ability to deliver reliable identity verification in high-throughput, mission-critical environments.
"Latent fingerprint identification is one of the most challenging problems in biometrics. Achieving the lowest FNIR at Rank-1 on the largest dataset in ELFT reflects years of focused algorithmic innovation. Our goal is not just to perform well in controlled settings, but to deliver consistent, high-confidence matching under real operational conditions," said Dr. Joshua Engelsma, Principal Scientist at ROC.
NIST's largest latent fingerprint benchmark uses the DoD-provided dataset, comprised of 5,259 latent probes. In this context, ROC's algorithms outperformed industry constituents delivering the highest Rank-1 accuracy, fastest search speeds, and smallest template size. This performance, combined with strong search efficiency relative to legacy technology providers, underscores the Company's competitive positioning in large-scale, mission-critical ABIS deployments and next-generation identity programs.
NIST benchmarks are widely referenced in federal, defense, and international agency vendor evaluations. ELFT rankings provide a transparent, standardized basis for comparing competing critical identity and intelligence technologies. ROC's unified Vision AI platform presents a significant competitive advantage by delivering cross-operational value to the end customer through efficiency gains, while supporting durable recurring revenue and attractive margins for the Company at scale.
"Performance at scale demands more than just accuracy; it requires a critical balance of precision and efficiency. Our algorithms are optimized to deliver consistent, low-latency performance even across millions of records," commented Keyur Patel, ROC's Director of Machine Learning.
The NIST ELFT program is widely regarded as the most rigorous independent benchmark for latent fingerprint matching and serves as a key reference point in agency evaluation and procurement processes worldwide.
Read ROC's Blog Post and Full NIST Scores here.
About ROC
ROC is a leading U.S. developer and manufacturer of Vision AI, delivering sovereign biometrics, video analytics, and mission intelligence through a unified platform. This enables agency and integrator partners to unlock faster, more accurate, and cost-efficient capabilities. At its core, ROC transforms raw pixels into real-time operational awareness for defense, public safety, and digital commerce. The Company is headquartered in Denver, Colo., with additional hubs in Grand Rapids, Mich., and Morgantown, W.V. For more information, please visit the Company's website: www.roc.ai.
Media inquiries:
Matt Aitken, VP of Marketing
[email protected]
Investor inquiries:
CORE IR
[email protected]
1 Rank-1 identification rate is what percentage of queries had their mate at the first retrieval rank (i.e., the closest match or in other words, the candidate with the highest similarity score to the probe image out of the entire gallery of candidates). Source: https://fiswg.org/
