Rank One says latest NIST test results show its consistent face biometrics accuracy and efficiency
Rank One Computing says it is the only “Western-friendly” submission in the recent Face Recognition Vendor Test (FRVT) from the National Institute of Standards and Technology (NIST) to receive top marks in both accuracy and efficiency.
In addition to demonstrating better than average accuracy performance across the board, Rank One had efficiency between three times and twenty times better than other prominent solutions, according to a company announcement.
Every other Western-friendly vendor submitting algorithms to the world-leading benchmark had multiple categories with below-average performance, according to Rank One. Rank One was in the second quintile in the “Wild” category, and the first in all others, out of the 221 face recognition algorithms submitted by 150 organizations, Rank One CEO Brendan Klare writes in a blog post.
Seven of the top 10 performers on the overall leaderboard are submissions from China.
The blog post shows Rank One’s favorable false non-match rate in each category compared to the median, its efficiency in size, resource usage and speed. The company also notes in the post that several prominent vendors have not submitted their algorithms for verification benchmarking in three or more years, and therefore should be avoided.
The post also breaks down the results demographically, showing that judged by the worst performing demographic cohort for each vendor, Rank One had the 12th best accuracy, exhibiting minimal impact due to race or gender, with higher accuracy scores for Black faces than White ones.
Rank One states that its combination of accuracy and efficiency make it effective for the development of any biometric face recognition system, from mobile and edge uses through enterprise and cloud deployment.
The algorithm evaluated in the NIST test is the one included in the ROC SDK v1.22, which was upgraded by the company earlier this year.
The company says the overall results show why it used by many organizations including in U.S. national defense, law enforcement, national banks and other industries.