Paravision biometric facial recognition software makes further accuracy gains with update

The fourth generation of Paravision’s biometric facial recognition has been released with a 36 percent reduction in error rate from an algorithm that had already placed among the leaders in NIST testing.

“Our team, our technology, and our process all allow us to rapidly iterate and improve our products, delivering meaningful advances in face recognition performance on a consistent basis,” says Charlie Rice, Paravision’s CTO.

Paravision currently sits third among developers of 1:N facial recognition algorithms, and finished second in the recent test for face biometrics with masked faces. The 36 percent decrease in error rates is in comparison to the algorithms evaluated in those tests.

Customers can implement Paravision’s facial recognition through SDKs or a cloud service, with optimization for leading platforms and operating systems from the server side to the network edge, according to the announcement.

Asked about the combination of biometric accuracy performance and flexible implementation, Paravision Chief Product Officer Joey Pritikin tells Biometric Update in an email that the company works closely with the world’s leading semiconductor companies to ensure optimal performance and compatibility across a wide range of operating environments.

“In particular, our face recognition toolkit has been optimized for a number of neural network inference acceleration toolkits, including NVIDIA TensorRT and Intel OpenVINO,” Pritikin explains. “So, for instance, we’re able to deliver high-speed face detection, image quality analysis, presentation attack detection (anti-spoofing), and 1:N face recognition on NVIDIA’s edge processing platforms, from Jetson Nano to Xavier NX. We are also hard at work on some very exciting next-generation, ultra-low power edge processing platforms, and will be sharing details on that in the weeks to come. In general, we’re big believers in GPUs and other AI accelerators due to the combination of power efficiency and price-performance, which is very compelling relative to traditional CPUs.”

In August, Paravision announced plans to shut down cloud photo storage app Ever, and that it would delete all images uploaded to the service as of the end of that month. Having done so, the company did not draw on that data to further train its algorithm.

“Computer vision is evolving at a dramatic pace, and we’re truly energized to harness the latest advances in AI and deep learning to help our partners solve some of the most challenging problems in identity, safety, and security,” states Paravision CEO Doug Aley in the announcement. “As a partner-focused, U.S.-based company delivering world-class accuracy, we feel we have a unique role to play in this regard.”

Article Topics

accuracy  |  biometric identification  |  biometrics  |  biometrics at the edge  |  cloud services  |  computer vision  |  facial recognition  |  Paravision  |  research and development  |  SDK