Paravision facial recognition SDK supports Rockchip edge AI processors
Paravision has revealed the optimization of its facial recognition SDK for Rockchip edge AI processors. This fine-tuning will expand the capabilities of its solution, allowing for broader support in IoT and access control applications.
Prior to this development, Paravision supported its solution for other processor manufacturers, including Ambarella and Hailo. The company claims that this solution offers a cost-effective entry point for new markets and applications, thereby promoting the adoption of advanced facial recognition technologies.
The SDK is compatible with various other Intel, Nvidia, Apple Silicon, and Arm processors. They leverage computer vision frameworks such as OpenVINO, TensorRT, CVflow, and CoreML.
“Our optimization for Rockchip not only broadens the applicability of our solutions but also simplifies integration, allowing our partners to easily implement top-tier face recognition technologies on the SoCs they already utilize,” says Joey Pritikin, chief product officer of Paravision.
Paravision has designed the facial recognition SDK modularly, supporting a wide range of hardware platforms, operating systems, and deployment scenarios. The software includes several functionalities, such as face detection, landmark identification, embedding computation, and template matching.
“Rockchip’s chipsets provide a robust foundation that allows our AI models to excel in varied and challenging environments and requirements, delivering reliable and highly accurate results across use cases,” says Charlie Rice, chief technology officer of Paravision.
Earlier this year, Innovatrics supported Rockchip processors by integrating its universal facial recognition SDK, the SmartFace Embedded Toolkit. “This integration allows for streamlined iris recognition processes directly on OEM and edge devices, using Rockchip RK3566,” says the firm.
Article Topics
AI chips | biometrics | biometrics at the edge | edge AI | facial recognition | Paravision | Rockchip | SDK
Comments