FaceTec claims best-ever face biometric authentication accuracy with new model
FaceTec has a new neural network model that it says delivers authentication accuracy with 3D face biometrics 876.56 percent better than its previous algorithm, and an even more massive improvement compared to the leading algorithm reported in the industry-leading benchmark.
The Face Recognition Vendor Test (FRVT) from the U.S. National Institute of Standards and Technology (NIST) does not test 3D data, and therefore does not include FaceTec’s algorithms. The developer argues that NIST’s results are irrelevant as 2D facial authentication is not sufficiently accurate to provide robust security, and has not been for five years.
The new algorithm developed by FaceTec, however, achieves a 1-in-125 million false acceptance rate (FAR), according to the announcement. The ramifications of this are described by FaceTec as “game-changing,” particularly in combination with liveness detection with 99.999 percent confidence.
“Our algorithms utilize the unique shape of the user’s 3D Face, not just 2D face data, and the results are exceptional,” states Josh Rose, FaceTec’s CTO. “When we compare them, our latest 3D face matching algorithm is approximately 8,600 percent better than the #1 NIST FRVT 2D algorithm.”
FaceTec has turned to the UK’s Age Check Certification Services (A.C.C.S.), which is accredited by the United Kingdom Accreditation Service (UKAS), which will perform N:N testing using the company’s 3D face biometrics.
“We are committed to the certification of the latest, state-of-the-art biometrics, age estimation, and identity verification systems, and to that end, we are offering the first testing framework for 3D Face Matching accuracy, which will enable next-generation biometric modalities, like FaceTec’s, to be evaluated at scale,” says Tony Allen, CEO of A.C.C.S.
Like all FaceTec algorithms, the new model utilizes a short video selfie, captured on any modern smart device or webcam, to perform both biometric matching and presentation attack detection.
“FaceTec’s 3D FaceMaps are created using standard 2D cameras but contain much more data than a flattened, 2D photo ever can,” comments Kevin Alan Tussy, CEO of FaceTec. “Think of the copper phone lines used for both dial-up modems and DSL internet. Just as the latter achieves vastly better results with the same hardware, similarly our patented innovations collect 3D data from ubiquitous 2D cameras and the results significantly outperform 2D face matching.”
This post was updated at 11:53am Eastern on May 4, 2022 to clarify which algorithms are included in each comparison.