Innovatrics is bringing its SmartFace facial recognition platform to Ambarella processors. The development will make it easier for equipment manufacturers to build edge devices with facial recognition capabilities.
The SmartFace Embedded solution can be deployed on the Ambarella CV22 and CV25 System-on-Chip (SoC) processors, both of which have been optimized for image and video-based facial recognition applications. Devices with the former need only 11 milliseconds to detect a face with a built in camera. Devices with the CV25 chip are slightly slower, but can still accomplish the same task in as little as 16 milliseconds.
According to Innovatrics, SmartFace Embedded is well-suited to access control, since facial recognition is contactless and is therefore more hygienic than legacy credentials. Cameras with the Ambarella chips will also be able to extract and process facial templates locally. That means that that data does not need to be sent to a central location, which in turn minimizes the amount of bandwidth and server space that organizations need to dedicate to access control.
“With a long-term track record of porting our algorithm to various hardware, we are now able to pre-process video streams directly on a smart camera,” explained Innovatrics SmartFace Technical Director Michal Vilagi.
The announcement reflects Innovatrics’ recent interest in edge technology. The company has already optimized SmartFace for Linux and NVIDIA Jetson systems, and has since gone on to release an updated SmartFace 5 version of the platform that takes advantage of edge processing to allow organizations to run security networks with an unlimited number of cameras.
Those interested in learning more about SmartFace can speak to an Innovatrics representative at next week’s ISC West conference at the Venetian Expo in Las Vegas. AnyVision and Paravision are some of the other facial recognition developers that have brought their solutions to Ambarella edge devices.