Skip to content
OptiExacta Labs

NIST FRVT · 1:N identification

Measured by NIST. 5th worldwide.

OptiExacta Facial Recognition is the first Indian algorithm to break into the NIST FRVT global Top 5 — placing 5th for webcam identification, and 6th for frontal mugshots.

#5Webcam Mugshots
#6Frontal Mugshots
#14Immigration Visa-Kiosk
#24Immigration Visa-Border

Validation

Ranked among the world's best, by the benchmark the world trusts.

NIST FRVT is the gold-standard benchmark for face recognition. OptiExacta placed 5th worldwide for webcam identification — the first Indian algorithm to reach the global Top 5.

First.

The first Indian algorithm to rank in the NIST FRVT global Top 5.

What FRVT is

Why this benchmark is the one that counts.

Anyone can claim accuracy. NIST is where it gets tested by someone else.

Independent, not self-reported

The US National Institute of Standards and Technology evaluates face-recognition algorithms on its own sequestered datasets. Vendors can't tune to the test — the results are an outside benchmark, published for anyone to check.

1:N identification, not 1:1

These results are 1:N identification: searching one captured face against a gallery of many enrolled identities, at a 0.3% false-positive identification rate. That's the hard problem behind finding a person in a crowd — not merely confirming a claimed identity.

Ranked against the whole field

Hundreds of algorithms enter each FRVT category, submitted by vendors and research labs worldwide. A rank here means OptiExacta was tested against that entire field, on NIST's own sequestered data — not a curated subset.

Performance

Fast enough, and small enough, to actually deploy.

The same evaluation measures how the engine behaves at scale — not just how often it's right.

~1.15s
Median template generation

Time to turn a captured face into a searchable template.

2,564 B
Template size

Small templates mean lower storage cost and faster transmission across a network.

1.6M+
Enrolled identities

Accuracy holds as the gallery scales past 1.6 million enrolled identities.

A benchmark is a starting point, not a deployment.

FRVT tells you the engine is good. Whether it works on your cameras, your lighting and your gallery is a different question — and the only way to answer it is to run it.

Run it on your footage