Trueface
Published in

Trueface

Trueface Achieves Top 10 Result in NIST Facial Recognition Vendor Test

We are proud to announce that of the 199 algorithms reported on the latest NIST Facial Recognition Vendor Test, 1:1, Trueface placed 7th in the category of genuine template comparison time.

For the uninitiated, the National Institute of Standards and Technology is a US government agency that evaluates advanced technologies for the benefit of the public market. The Facial Recognition Vendor Test (FRVT) is considered the most rigorous and difficult evaluation for facial recognition technology; clients often will not even consider a vendor unless they are NIST FRVT certified. This evaluation measures the performance of an algorithm on different datasets of varying control and quality such as Visa, Mugshots, and Images in the Wild. Moreover, algorithms must meet certain difficult requirements to even be considered for the evaluation including timing requirements (while running in single-thread mode), operating system and linking environment compliance, and most importantly, the ability to interface in C++. More information regarding the evaluation can be found here.

What is a genuine match and why is comparison time important?

When a facial recognition algorithm is provided with an image of a face, it will generate a template describing the face as an output. This template, an array of numbers of fixed length, mathematically describes certain attributes of a face such as the distance between eyes, the angle of the nose, etc. When two of these face templates are provided to the algorithm, it compares the templates to determine similarity, and thus the probability that the identity of the two templates is the same. Hence, in a real-time streaming environment, the time taken to make this template comparison is immensely important.

At scale in the real world, template match speed is critical in everyday application. Increased match speed allows workers’ identities to be verified seamlessly and provide quicker access to a worksite, resulting in more time spent on the job and not in the queue to get in, thus increasing productivity. Faster template match speed also allows security teams to identify people of interest like missing children or known threats much more quickly, making environments safer in the process. When the algorithm is presented with millions of potential matches per second, speed is essential, no matter the real-world application.

Why the sudden change in template comparison time?

A scrupulous follower of the NIST FRVT reports may have noticed Trueface’s template comparison time improved significantly in the latest report, despite not having made another submission. Curiously, our very own engineers noticed that our initial submission’s match times did not correlate with the results we had achieved on equivalent testing hardware and operating system and repeatedly contacted NIST several times to explain the difference. While NIST can neither confirm nor deny that Trueface’s own investigative reporting and flagging had anything to do with the updated speeds in their report, let’s just say we’ve stopped calling them.

Here at Trueface, we are continuously improving our facial recognition algorithm, and we anticipate that our performance will improve on the next NIST FRVT submission. If you are interested in learning how our technology can benefit your business please contact us at sales@trueface.ai.

Disclaimer: Results shown from NIST do not constitute an endorsement of any particular system, product, service, or company by NIST. For more information, visit https://www.nist.gov/programs-projects/face-recognition-vendor-test-frvt-ongoing

--

--

--

Transforming camera data into immediately actionable information with computer vision.

Recommended from Medium

Visual Automation: Boost Agent Productivity with Smart Eyes

The contract is dead

When Should Machines Make Decisions?

XLerated AI — Part V — Hollywood

Semantic Internet of Things

The brand new ItsAlive 2.0 ⚡ brings exciting features to your chatbots 🤖

AI is coming. Is your job secured?

When you should use Constraint Solvers instead of Machine Learning

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Cyrus Behroozi

Cyrus Behroozi

Senior computer vision software developer at Trueface.ai | cyrusbehr.com

More from Medium

What are the Biometric Techniques Used for Person Recognition?

C++ and WebRTC: Video and Audio to the Browser

Building an android application to control Tello drone flight and perform real-time object…

Object Detection with KotlinDL and Ktor