NEWS
FAICETECH'S LATEST NIST FRVT SCORES RELEASED
Independent Testing Confirms Industry-Leading Accuracy Across All Benchmarks
Independent Testing Confirms Industry-Leading Accuracy Across All Benchmarks
When it comes to facial recognition technology, independent testing is everything. The National Institute of Standards and Technology, part of the US government, runs a program called the Face Recognition Vendor Test (FRVT). It's the gold standard for testing facial recognition algorithms, and it's completely independent. Vendors like FaiceTech submit our algorithms to NIST, and their team puts them through millions of face comparisons across hugely diverse datasets. The vendors have no influence over how the testing happens or what the results show. It's about as close to completely unbiased testing as you can get. Think of it as the crash-test standard for facial recognition. Any vendor can claim accuracy numbers. NIST numbers come from neutral third parties running tests in a controlled, public way.
We're proud to share FaiceTech's latest NIST FRVT results, submitted March 2025 and published in the report card generated 30 March 2026. On the Visa-Border dataset — the benchmark most relevant to real-world identification — our algorithm achieved a false non-match rate (FNMR) of just 0.0023 at a false match rate of 0.000001. That's 99.77% verification accuracy at an extremely strict threshold. On the Mugshot-Mugshot dataset, FNMR was 0.0032 (99.68% accuracy), and on Border-Border comparisons, 0.0115 (98.85%). These aren't marketing claims. These are the actual results you'll find on the official NIST report card.
Full results are available at face.nist.gov/frte/reportcards/11/faicetech_000.html, where you can see the complete data, testing methodology, and all the details behind these benchmarks.
Here's the truth: any vendor can claim accuracy numbers. A company can put out a press release saying their algorithm is 99.99% accurate. But who tested it? Their own team, usually. Did they test it fairly? How do you know? With NIST, you get something different. An independent government agency runs the test. They use their own datasets, their own methodology, and they publish the results publicly. You can read them yourself. You can compare vendors side by side. When a vendor won't share their NIST results, ask why. That's a fair question. You wouldn't buy a car that hasn't been crash-tested. Why would you deploy facial recognition without independent testing?
Accuracy percentages might sound abstract, but they translate directly into real-world impact. A 99.99% accuracy rate means dramatically fewer false alerts, less wasted security time, and more reliable identification. For a large retail environment, a stadium, or an airport processing thousands of faces every day, the difference between 99% and 99.99% is massive. It means the difference between dozens of false alarms per day and almost none. It means your security team spends their time on genuine threats, not chasing ghosts. It means you catch the person you're actually looking for, not someone who just looks similar.
Let's address something important. Early facial recognition systems had real bias problems. The accuracy varied significantly across different skin tones, ages, and demographics. It was a serious issue, and it mattered. NIST specifically tests for this. Their testing includes diverse populations across multiple demographic groups. FaiceTech's results show consistent performance across demographic groups, with FNMR values on the Visa-Border dataset staying within a tight range across countries of birth, age brackets and gender groups. The report card includes a full breakdown by country of origin (from El Salvador to Nigeria to Japan), by age (under and over 45) and by gender — all publicly available for scrutiny. This isn't about checking a box. It's about building a system that's fair and reliable for everyone who uses it.
The NIST report card is public. You can visit face.nist.gov/frte/reportcards/11/faicetech_000.html right now and see the full technical results. We're confident in these benchmarks because they're independent, rigorous, and proven in the field. If you want to see how FaiceTech's accuracy translates to your specific use case, let's talk. Our team can walk you through the results, answer your technical questions, and show you how this performance works in real security deployments.