Select your language

The interface for running AI models on the NPU.

After compilation, the SDK generates images in the project/image/output/ folder, ready to be flashed via TFTP or USB. 5. AI Integration with the SigmaStar SDK

Before diving into the code, you need a stable cross-compilation environment. Prerequisites

Run the make command to compile the bootloader, kernel, and rootfs. make all Use code with caution.

Use the SigmaStar PC-based tool to convert an .onnx model to a .sim or .bin file.

In your C/C++ code, use the MI_IPU APIs to load the model, feed it buffers from the MI_VI module, and retrieve detection results. 6. Best Practices for Developers