Paravision unveils edge AI toolkits for facial recognition applications
Paravision has released new software development kits (SDKs) to facilitate the design and deployment of embedded facial recognition systems based on its edge AI (artificial intelligence) technology.
Edge AI solutions for facial recognition are already powering several devices on the market that are both ultra-low power and small form factor, as well as extremely fast and accurate.
The new toolkits released by Paravision now support several facial recognition functions, including face detection, landmark detection, image quality assessment, liveness and anti-spoofing, template creation, and 1:1 or 1:N face matching.
Commenting on the news, the computer vision firm said the above facial recognition functions can be integrated with backend systems in a hybrid architecture. In other words, a system built via the Edge AI SDK could potentially run face detection, quality metrics, liveness, and image cropping on the edge while enabling face matching itself in the cloud. Such an architecture could maximize responsiveness while using biometric matching galleries too large for on-device storage.
In terms of technical specifications, Paravision’s toolkits are compatible with a wide range of operating systems, including Embedded Linux with both Ambarella and Nvidia, Linux and Windows with Intel, and Android with ARM.
They also support various programming languages, namely C++ on Embedded Linux, Linux, and Windows, Python on Linux and Windows, and Kotlin on Android.
In addition, the SDKs enable development on various computer vision frameworks, such as Intel CPU (OpenVINO), Nvidia Jetson-family SOM (TensorRT), Ambarella (CVflow), and RM (PyTorch).
Created via a partnership with specialist systems integration provider Teknique, the new Reference Design Kits aim to enable rapid hardware and software development of Ambarella-based edge AI solutions.
“The software that we’ve created is outstanding, but the opportunity to give our partners a customer to rapidly, cost-effectively develop with very low risk is super-exciting,” Paravision CPO Joey Pritikin explained to Biometric Update in an interview on the Converge 2021 announcements.
Using Teknique’s Ambarella System-on-Module (SOM) boards and camera boards built with best-in-class CMOS image sensors, Paravision has already created solutions demonstrating the new capabilities of its new Edge AI SDK.
The release of the new SDK comes months after the company closed a $23 million funding round.