Neurotechnology Achieves Highest Score for Iris Recognition Accuracy in IREX 10 Rank1 Category
Neurotechnology's iris recognition algorithm submission to the NIST IREX 10 evaluation improves performance and surpasses the competition in the Rank1 category.
Vilnius, Lithuania – June 20, 2023 – Neurotechnology, a provider of deep-learning-based solutions and high-precision biometric identification technologies, announced today that the company's iris recognition algorithm has achieved the top position in the Rank1 category in the National Institute of Standards and Technology (NIST) Ongoing Evaluation of Iris Recognition (IREX 10). The latest algorithm outperformed other submissions in a single-eye assessment with a 99.14% accuracy rate and in a two-eye assessment with a 99.73% accuracy rate.
Rank-based accuracy relates to performance for investigational tasks and is mostly relative to adjudication scenarios where algorithms return a list of candidates to operators for further inspection. The Rank1 rate measures the frequency with which algorithms accurately identify the correct person within the top-ranked predictions, making it indispensable for real-time systems.
"We are excited to achieve the top accuracy in NIST's evaluation categories and are thrilled to see our algorithm's miss rate almost twice as low as the closest competitor in some assessments," said Evaldas Borcovas, biometric research lead at Neurotechnology. "I believe this submission to IREX 10 is one extra step toward our goal to develop overall the most accurate iris recognition algorithm."
IREX 10 Rank1 accuracy metrics can typically relate to real-time civil and criminal identification scenarios. For visitor management systems, this metric is crucial to ensure precise identification of a visitor and further granting attributed permissions.
For law enforcement and criminal investigation systems, the Rank1 metric can showcase how effective the algorithm is in displaying the correct suspect in the first place of candidates – saving officers time and reducing the need for additional searches.
Neurotechnology's latest submission also showed improvements across other IREX 10 assessments. Measuring accuracy at FNIR@FPIR 0.01, NIST rated all the algorithms to the leaderboard, placing Neurotechnology at 6th place with two-eyes and 5th place with single-eye, among 29 participants. FNIR@FPIR shows a tradeoff between false negative identification versus false positive identification at some operational threshold – in this case, 1 / 100 false positive identification.
Neurotechnology, as a multi-biometric solutions developer, is also an active participant in other NIST evaluations of different biometric modalities. Over the years, the company's algorithms have been among the top performers in evaluations including MINEX III, PFT III, FRVT, ELFT, and SlapSeg III.
Neurotechnology is a developer of high-precision algorithms and software based on deep neural networks and other AI-related technologies. The company was launched in 1990 in Vilnius, Lithuania, with the key idea of leveraging neural network capabilities for various applications, such as biometric person identification, computer vision, robotics, and artificial intelligence. The company's solutions and products have been used in more than 140 countries worldwide and in many national-scale projects for national ID, passports, elections, and border control, including India's Aadhaar program, the Ghana General Elections, the Democratic Republic of the Congo Voter Deduplication and other projects that collectively process the biometric data of nearly two billion people.
Jennifer Allen Newton
Bluehouse Consulting Group, Inc.
jennifer (at) bluehousecg (dot) com