Paravision develops biometrics and detection for faces with masks and social distancing

Paravision has been working on a toolset to extend its computer vision offerings beyond biometric facial recognition, to provide enhanced identification, safety and security in the context of the COVID-19 pandemic, according to a LinkedIn post by CEO Doug Aley.

In response to the unique challenges posed by COVID-19, Paravision rapidly developed proof-of-concept tools with computer vision to support a variety of applications.

The expanded capabilities include advanced object detection, tracking, and activity recognition, according to a video posted to LinkedIn. The company plans to roll out the toolset throughout 2020. The video shows the mask detection and identification capability, including in crowds, with an interesting moment showing a busker playing a saxophone, who is judged to be wearing a mask by the system while turned away, but whose lack of a mask is detected when he turns towards the camera. Social distancing metrics, detecting individuals standing within approximately 2 meters of each other, are also demonstrated in the video.

Chief Product Officer Joey Pritikin lauds the biometric accuracy performance of the company’s out-of-the-box technology for faces partially occluded by masks.

“For those who will surely ask: Based on the newness of the challenge, we don’t yet have a statistical basis for assertion of accuracy with masks on. The limiting factor is the dataset to do the benchmarking,” Pritikin notes in a LinkedIn post, and invites collaboration to work on that limiting factor.

A number of companies have recently announced facial recognition algorithms tuned for recognition of people wearing masks.

The company claims the new capabilities were all developed in the first week of April.

Paravision recently announced it was the only biometrics provider to be ranked in the top five of both NIST FRVT reports, 1:1 and 1:N, for accuracy.

Article Topics

biometric identification  |  biometrics  |  computer vision  |  facial recognition  |  mask detection  |  Paravision  |  research and development