In the latest NIST Face Recognition Vendor Test (FRTE), ROC delivered one of the world’s most consistently accurate face recognition algorithms, achieving top marks across 1:1 Verification, 1:N Identification, and 1:N Investigation—ranking among the most accurate globally and standing as the leading American-made solution for mission-critical applications in border verification, massive investigative searches, and high-throughput traveler scenarios.
ROC, the only American-made multimodal biometrics and Vision AI provider, today announced record-setting results in the NIST Face Recognition Vendor Test (FRTE). These results demonstrate more than technical superiority—they confirm ROC’s role as a strategic provider of trusted identity technology at national scale. With algorithms tested against 400+ global submissions and measured across 24 diverse datasets, ROC’s performance validates its leadership in trusted, American-made biometrics built for defense, public safety, and digital trust to deliver real-time decision advantage to U.S. agencies and allies.
ROC’s new algorithm achieved one of the most accurate face recognition results the world has ever seen, delivering across-the-board excellence in accuracy, efficiency, and speed.
#1 Western Vendor for Investigative Search (1:N)
In NIST’s 1:N Investigative Search benchmark, matching immigration lane and registered traveler kiosk photos against a database of 1.6 million identities, ROC ranked #1 Western vendor in investigative search accuracy, excelling at both datasets. Immigration Lane tests simulate rapid identity checks against massive watchlists, while Traveler Kiosk tests reflect real-world challenges of lower-quality or off-angle images.
Rank-1 miss rates on investigative searches on a visa applicant photograph database using images from traveller kiosks and immigration lanes, compared to Western vendors in the NIST FRTE 1:N Investigative Search benchmark. Database size 1.6M persons.
#1 Western Vendor for Visa-Border & Border-Border Verification (1:1)
In NIST’s 1:1 Verification benchmark, ROC ranked #1 Western vendor and #3 globally. This test measures how accurately algorithms verify whether two face images belong to the same individual, an essential capability for border security and high-assurance identity verification. Visa-Border involves matching visa application photos against live border captures, while Border-Border involves matching live border captures against prior border images.
False Non-Match Rates (FNMR) at a False Match Rate of 1 in 1,000,000 for the Visa-Border and Border-Border identity verification datasets, compared to Western vendors in the NIST FRTE 1:1 Verification benchmark.
#2 Western Provider for 1:N Identification & Off-Pose Performance
In NIST’s 1:N Identification benchmark, ROC stood out for its best-in-class efficiency and robustness on challenging face images. On profile (90-degree rotation) photos, ROC achieved the second-lowest error rate. At the same time, ROC delivered the fastest search speeds of any vendor, searching a 12-million-record facial database in just 1.45 seconds using modest hardware resources, proving unmatched scalability for national-level deployments.
Search speeds and identification error rates on profile imagery (faces rotated 90° from the camera) compared with Western vendors in the NIST FRTE 1:N Identification benchmark.
“NIST FRTE is the most rigorous and respected global benchmark for face recognition. This achievement reflects more than a decade of focused research and development here in the United States. What sets this algorithm apart is its consistency. It performs at the highest level across verification, identification, and investigative search. This consistency is what makes it among the world’s best, and why our partners trust ROC for their missions.”
– Dr. Brendan Klare, Chief Scientist and Co-Founder, ROC
“Our NIST performance isn’t just about a breakthrough algorithm, it’s a testament to our unique development process. ROC algorithms are built on a foundation of decades of facial recognition expertise, continually advanced with the latest in AI research and optimization. That’s how we deliver solutions that perform with high accuracy across every scenario, from one-to-one verification to large-scale investigative searches. At ROC, we’re committed to pushing the boundaries of face recognition to achieve what was once thought impossible.”
–Keyur Patel, Director of Machine Learning, ROC
These results confirm what matters most: consistency. Across 1:1 verification, 1:N investigation, and 1:N identification, ROC algorithms perform with the same high accuracy, speed, and reliability in every scenario, making the company one of the most consistent and capable Western providers in face recognition today.
About NIST FRTE
The NIST Face Recognition Technology Evaluation (FRTE) is the national benchmark for face recognition systems. It evaluates algorithms in identity verification scenarios, such as 1:1 matching and 1:N identification, under realistic operating conditions using large controlled datasets. It evolved from the earlier Face Recognition Vendor Test (FRVT) to better reflect today’s technology landscape and includes a broader range of face identity evaluation tracks. By publishing detailed leaderboards and reports, NIST provides organizations and integrators with transparent, objective data to guide the responsible deployment of face recognition technology.