CyberLink’s AI facial recognition engine scores in top 20 in NIST biometric accuracy testing

CyberLink’s AI biometric facial recognition engine has scored a 99.7 percent accuracy rate on the NIST FRVT 1:1 (Facial Recognition Vender Test), the company announced.

According to the test results published in March, FaceMe’s algorithm had a 99.70 percent accuracy rate in VISA 1E-4 and 97.17 percent in WILD 1E-4 precision assessments, the latter ranking 18th globally. The company sits 19th on the overall leader board.

“Since its release, CyberLink’s FaceMe has consistently achieved high rankings in leading worldwide industry assessments,” said Dr. Jau Huang, CEO of CyberLink, in a prepared statement. “FaceMe is seeing strong improvement compared to our previous NIST benchmarks, up 45 percent in accuracy rate, anchoring its prominent ranking in all testing criteria. Thanks to our strong precision standings and unparalleled integration flexibility across edge devices, FaceMe continues to be the best fit for AIoT deployment.”

FaceMe leverages an edge-based architecture for processing and security, supporting over 10 operating systems, including Windows, Android, iOS, and various Linux distributions such as Ubuntu x86, Ubuntu ARM, RedHat, CentOS, Yocto, Debian and JetPack. As the engine operates on low-power CPU, facial recognition can be rolled out on IoT and AIoT devices, high-end servers, workstations and GPU-equipped personal computers.

CyberLink’s biometric facial recognition engine was integrated with Altek edge AI box to develop intelligent edge computer vision for AI security control solutions.

Article Topics

accuracy  |  biometric testing  |  biometrics  |  CyberLink  |  Face Recognition Vendor Test (FRVT)  |  facial recognition  |  NIST