Create Your Own Model

KNEO Pi empowers developers to explore and deploy AI models for a wide range of applications, such as object detection, image segmentation, and more. By porting pre-trained models or creating custom ones, you can take full advantage of the Neural Processing Unit (NPU) on the KNEO Pi for optimized AI performance.

Porting a model to the KNEO Pi involves converting it to a format compatible with the device's NPU. This requires the use of the model toolchain, a utility designed to adapt and optimize models for the KNEO Pi. The toolchain handles tasks such as quantization and compiling the model into an instruction set that the NPU can execute efficiently.

Info

For detailed steps on using the model toolchain, refer to Advanced - Kneron Model Toolchain Guide.

Explore Pre-Trained Models

You can explore a wide range of pre-trained models hosted on GitHub and other repositories to kickstart your AI projects. These models provide a solid foundation and can be customized or optimized to fit your specific application needs.

Model Porting Examples

To help you get started, we provide two detailed step-by-step porting examples:

• YOLOv7 for Object Detection: An object detection model.
• PIDNet for Segmentation: A popular semantic segmentation model.