Paravision tops Western facial recognition providers in NIST FRVT 1:1

San Francisco-based Paravision says its fifth generation facial recognition technology was ranked as the top performing biometric algorithm submitted from the U.S., UK and EU, placing the company behind only 4 providers globally in the NIST’s newest Face Recognition Vendor Test (FRVT) 1:1 report, published February 23, 2022.

The company states in a press release that it has shown improvements in accuracy throughout its submissions since 2018. Paravision points to a 65 percent decrease in technology error rates in 2021’s Visa/Border test. Overall, Paravision claims that performance increased 85 percent across all tests since its first submission in 2018.

Paravision also says progress on facial recognition on demographic performance across gender, nationality, and age was made. The company claims that across gender, age and nationality demographics, its face recognition now shows less than 0.65 percent false non-match rates at 1 in 100,000 false match rates for every demographic group tested.”

The report continues Paravision’s positive showings in NIST FRVT 1:1 tests. In January, the fifth generation of facial recognition showed a 70 percent reduction in error rates compared to its previous submission, and it was named top-five globally and number one in the U.S. in a 2021 analysis.

Charlie Rice, Paravision’s CTO, says, “Our machine learning and engineering teams are truly remarkable and passionate about continuously progressing our technology and improving accuracy, and Paravision’s performance in the NIST FRVT reports showcases that. We strive to innovate on behalf of our clients, and our rate of improvement is a reflection of this core value.”

Entries from two Russian developers and a pair of algorithms from China’s SenseTime are the only ones listed above Paravision in the February 23 leaderboard update.

Article Topics

accuracy  |  biometric testing  |  biometrics  |  Face Recognition Vendor Test (FRVT)  |  facial recognition  |  identity verification  |  NIST  |  Paravision