Innovatrics and Blaize are joining forces to improve facial recognition at the edge. To that end, the two companies will pair Innovatrics’ SmartFace Embedded facial recognition algorithm with Blaize’s Pathfinder P1600 Embedded System on Module (SoM) for edge devices.
The joint solution turns a regular security camera into a smart camera that can perform select biometric functions locally on-device. The SmartFace Embedded system can detect landmarks and extract a face template in less than a second, enabling the rapid match of a face to an image stored in the cloud.
The goal is to minimize the amount of data that needs to be transferred to a dedicated data center. Since the Innovatrics and Blaize offering pre-processes more footage on-device, it reduces the computational strain on a broader network and makes it easier for an organization to scale up its security operations. In that regard, organizations can select the specific biometric functions that they are interested in when deploying SmartFace to further reduce the data footprint and realize even greater efficiency benefits.
The Innovatrics and Blaize solution is designed primarily for access control and public security applications. Each camera outfitted with the SmartFace Embedded and the Blaize SoM can process multiple video streams, and offers plug-and-play utility that makes it easy to add new devices to an existing network.
“Blaize focuses on delivering low power and low latency AI inference solutions at the edge, and Innovatrics with their SmartFace Embedded enables customers to realize the value of AI in smart city, smart retail, safety and security applications,” said Blaize Marketing Head Barrie Mullins. “Innovatrics facial recognition technology seamlessly integrates into our cost-effective and flexible edge form factor solutions.”
Innovatrics believes that the new offering will be particularly appealing to organizations that have struggled to find solutions while GPUs are in short supply internationally. The company’s SmartFace solution also boasts passive liveness detection to prevent spoofing.
May 16, 2022 – by Eric Weiss