Paravision is celebrating its strong performance in the NIST’s latest Face Recognition Vendor Test (FRVT) 1:N Identification report. The 1:N test evaluates facial recognition algorithms based on their ability to match an unknown face to a known identity.
This time around, Paravision’s solution was measured against more than 200 facial recognition algorithms from 60 different vendors. Paravision excelled despite the tough competition, achieving top-three scores in multiple face matching categories, including mug shots, webcam photos, and aging faces over a period of 18 years. In each case, Paravision had the top-ranked solution in the US, Europe, and the UK. It was also the top-ranked solution in the world when matching profiles turned 90 degrees from the camera.
The results make Paravision the only company to achieve a top-5 score on both the 1:N and 1:1 Verification tests. The company claimed the top spot overall when the 1:1 results were released in August, after previously posting the best score in the Wild Images category as Ever AI. The San Francisco-based company rebranded as Paravision in the second half of 2019.
According to Paravision, the results speak to the progress the company has made in the past few years. The company has jumped from the top 10 to the top three (while reducing its false negative identification rate) since it began submitting its algorithm to the NIST program.
In other news, Paravision Chief Product Officer Joey Pritikin recently participated in a virtual panel discussion that explored the issues of convenience, security, and privacy in the biometrics industry. The podcast discussion is now available through the FindBiometrics ID Talk channel, and was presented in collaboration with ISC West in an effort to advance the security conversation during the COVID-19 pandemic.
March 26, 2020 – by Eric Weiss