Ceva and embedUR systems have collaborated to introduce the ModelNova platform for Ceva’s NeuPro NPU series, with the goal of broadening the edge AI model landscape.
ModelNova offers ready-to-use AI models for ultra-low-power NeuPro-Nano NPUs and high-performance NeuPro-M NPUs, catering to various applications such as wearables, smart home devices, robotics, and AR/VR.
The platform is equipped with pre-built models, datasets, and use-case blueprints, streamlining the process of prototyping and deploying AI features.
Elia Shenberger, senior director of product marketing, AI Division at Ceva, commented, “ModelNova serves as a valuable extension of our NeuPro ecosystem, granting developers access to a diverse library of optimized AI models. Our collaboration with embedUR empowers customers to create intelligent edge devices with minimal development effort.”
These models are deployed and fine-tuned using TensorFlow Lite Micro, seamlessly integrated with Ceva’s NeuPro Studio SDK for applications like object detection, face recognition, and anomaly detection.
The emphasis lies on “physical AI,” injecting intelligence directly into devices to enable autonomous operation without heavy reliance on cloud connectivity.
Ceva’s versatile NeuPro NPU family caters to the low-power and high-performance demands of a wide spectrum of edge AI applications, spanning from basic to advanced connected devices.
ModelNova’s offerings are readily accessible, with new models and use cases continually being added.
Earlier this year, Ceva forged a partnership with ALi Corporation to integrate Ceva’s NPUs into ALi’s Video Display Sub-System (VDSS) platform.
Related
AI/ML | Ceva | device intelligence | edge AI | edge computing | embedded systems | embedUR | NPU
