Kakao shows biometric face verification accuracy with strong FRVT result

South Korea-based artificial intelligence developer Kakao Enterprise Corp. has burst into the top performers for biometric accuracy of the ongoing 1:1 Face Recognition Vendor Test (FRVT) from the U.S. National Institute of Standards and Technology.

The latest update from NIST, issued in late-June, shows the ‘Kakao-008’ algorithm scored the highest accuracy rate for matching photos in the kiosk category, with an error rate of only 0.0417. The company also recorded top 5 results with mugshot photos, border and ‘visaborder’ photos, and ranked in the top 10 for biometric accuracy with visa photos and mugshot photos aged 12 years.

Kakao has participated in the NIST FRVT since 2019, but first submitted the 008 algorithm in May. Kakao Enterprise is a subsidiary of Kakao Group, which also operates Kakao Pay and Kakao Brain, the former of which has also submitted an algorithm to the FRVT. The same corporate group also includes a messaging service that launched its own digital identity authentication platform in 2017.

The Kakao Enterprise subsidiary was founded in 2019, according to Crunchbase, and has raised $92 million in funding so far, giving it unicorn status as a billion-dollar valuated startup.

The Korea Bizwire reports that Kakao is now refining its facial recognition algorithms with a CCTV system set up to enable biometric payments at an unstaffed convenience store within its head office.

Algorithms from Chinese facial recognition developers Cloudwalk, SenseTime and Megvii were found among the top-performing entrants for biometric accuracy, as previously.

Article Topics

accuracy  |  algorithms  |  biometric testing  |  biometrics  |  face biometrics  |  Face Recognition Vendor Test (FRVT)  |  Kakao  |  NIST  |  South Korea  |  startup