Neurotechnology, a provider of deep learning-based solutions and high-precision biometric identification technologies, today announced that the company’s latest face recognition algorithm showed significant improvements among the top NIST FRVT testing results released on January 13, 2022.

The Face Recognition Vendor Test (FRVT) conducted by the National Institute of Standards and Technology (NIST) is the most reliable series of large-scale, independent evaluations for face recognition algorithms in verification (1:1) and identification (1:N) scenarios. Immense datasets containing photos of faces are used during the evaluation to measure the performance of face recognition algorithms developed worldwide.

The new face recognition algorithm from Neurotechnology has demonstrated significant advancement in both FRVT 1:1 and FRVT 1:N NIST testing, showing comprehensive performance across identification and verification testing scenarios.

“Consistency and dedication are crucial to our sustained R&D accomplishments,” said Evaldas Borcovas, biometric research lead at Neurotechnology. “Previously our team achieved the best algorithm accuracy in fingerprint recognition evaluations, and now we are seeking to do the same in face recognition evaluations. Based on our experience, and these latest algorithm results, I am confident that we are moving in the right direction.”

Neurotechnology - FRVT 1x1 PFT.PNG

In the FRVT 1:1 Verification evaluations, Neurotechnology’s face recognition algorithm showed significant performance improvements, including:

  • In the top 3% of most accurate results for border control supervised (Visa Border, Border) and unsupervised (Kiosk) scenarios among 702 submissions by 255 providers.
  • Among the top 3% of algorithms for accuracy with masks from a total of 319 entries.

Nuerotechnology - FRVT 1xN PFT.PNG

The face recognition algorithm also showed significant performance improvements in the FRVT 1:N Identification evaluations, including:

  • In the top 4% of the leading results matching frontal and profile mugshots scenarios among 341 submissions by 93 different providers.
  • Top results among border control supervised (Visa vs Border, Border vs Border ΔT ≥ 10 YRS) and unsupervised (Visa vs Kiosk) scenarios.
  • Leading score by template size. Considering the template size, the algorithm showed the best results among all other submissions with the same template size.

These results in the NIST FRVT demonstrate that the latest face recognition algorithm from Neurotechnology continues the company’s strong track record of providing face recognition products that are among the top performing solutions for some of the most common situations in civilian and law enforcement scenarios, as well as offering industry-leading efficiency by template size, extraction, and matching speed performance.