Paravision is trying to make it easier for businesses to implement facial recognition at scale, and for developers to integrate facial recognition into their own applications. To that end, the company has released a slew of new features and product updates that cover a number of different use cases for Paravision’s technology.
The most noteworthy new feature is Paravision’s Scaled Vector Search (SVS) tool, which allows organizations to perform matches with a larger database. The SVS engine can process hundreds of transactions every second with minimal latency, and generates accurate results in databases with tens of millions of unique identities.
The Streaming Container 5, meanwhile, is similarly being released to support large scale deployments. The solution can track the same face across multiple video feeds, even when there are many other faces present in each frame. Streaming Container 5 can handle any number of video streams, though it is capped at 250 frames per second.
On the mobile front, Paravision has released a new Face SDK for Android to help developers who want to build apps with facial recognition capabilities. The Android SDK supports 1-to-1 and 1-to-N use cases, and includes reference apps and user interface recommendations to help speed up the development process.
Many of the other updates will make Paravision’s technology more readily compatible with the Ambarella CVflow platform. The company started supporting the platform late last year, and is now releasing an enhanced Face SDK specifically for smart cameras that run on a CVflow chipset. Those cameras can run at the edge, and can now process as many as 40 frames per second thanks to the new SDK. Paravision has also released a new Liveness and Anti-spoofing SDK that leverages Ambarella’s visible light, near-infrared, and depth-sensing technology to identify spoofing attempts and provide users with an additional layer of security.
Paravision itself has formed a partnership with Teknique, which specializes in development for the Ambarella platform. The two companies are now working together to create a collection of hardware and software reference designs for edge facial recognition.
Finally, Paravision is moving beyond pure facial recognition with the launch of a new computer vision SDK. The SDK can enable new applications like object detection and people counting, which can help enforce occupancy limits or prevent tailgating at a point of entry.
The Paravision portfolio provides coverage for cloud and on-premises deployments, and has potential applications in industries ranging from payments to access control. The company brought in $23 million in its latest funding round, and it is consistently one of the top-performing developers in independent NIST evaluations.
October 1, 2021 – by Eric Weiss