AnyVision AI facial recognition software available on Ambarella chips

AnyVision, a provider of AI software for recognition, said its software is now available on Ambarella’s CV2x family of edge AI vision SoCs. The software will help extract actionable data from video feeds for biometric identification capabilities such as facial recognition, person detection, and other useful functions such as usage analytics.

The combination of AnyVision AI software with Ambarella’s chips pushes compute-intensive processes out to edge devices. Functions such as facial recognition can be performed on camera with increased performance and data security, while also reducing overall costs required by previous generation systems that required separate servers to do the processing. Another advantage—new applications that leverage facial recognition for payment processing are possible in markets such as retail stores that might now have had enough space or power to house those servers.

AnyVision’s Recognition AI adds capabilities to Ambarella’s chips via an open SDK for Ambarella’s CVflow platform. AnyVision executives expect the pairing to enable new applications including the following:

• Touchless access control: The pandemic has created a need to minimize physical contact between people and surfaces (doors, payment terminals, and such), and interest in touchless access control solutions is skyrocketing. AnyVision says its Recognition AI software uses deep neural networks to extract and compare the faces of those entering a facility with those registered in the system, providing a recognition accuracy rate of approximately 99 percent.

• Video Surveillance: As noted above, cameras with powerful chips and AI software can be used instead of the combination of costly on-premise servers and older generations of low-resolution video cameras. AnyVision says cameras themselves can become powerful edge computing devices with embedded facial recognition and watchlist-alert functionality, even in environments such as large crowds or low light environments.

• Liveness Detection: AnyVision said OEMs using edge AI technology can use 2D and high-resolution 3D liveness detection to prevent unauthorized access by spoofing (using a 2D picture held up to the camera, for instance) in order to bypass the access control or video surveillance system.

Ambarella is one of the leading providers of chips for computer vision applications such as security cameras. The CV2x chip family offers up to 4K resolution and 60fps video streams. The company recently introduced new chips that support up to 8K resolution or multiple 4K streams on a single chip. For AnyVision, that could eventually be a path for continued expansion as occupancy monitoring, missing persons identification, and other applications benefit from the enhanced accuracy provided by higher resolution imaging.

Article Topics

access control  |  AI chips  |  Ambarella  |  Anyvision  |  biometric identification  |  biometric liveness detection  |  biometrics  |  biometrics at the edge  |  facial recognition  |  video surveillance