Lightning Talk: Zephyr Portability with an AI Application on Very Different MCUs
summary
Single AI face detection app runs unchanged source on two very different MCUs by swapping DeviceTree + Kconfig entries only the target boards differ the platforms are a Cortex M7 RT1060 doing inference in software around 100 ms per image and an M33‑based M6N using NXP NPU accelerated TensorFlow Lite models at about 20 ms per image camera and display interfaces differ and a limited frame buffer forces display driver changes which NXP contributed back working demo and code are on GitHub and slides are public
AI-written summary · about
related talks
10mLightning Talk: Debugging Edge AI on Zephyr and Lessons Learned
18mLightning Talk: If Zephyr Wants To Power AI Cameras, What Must Change?
36mTinyML at the Edge: Deploying and Optimizing AI Workloads on Zephyr RTOS
42mPorting an AI Powered Wearable Health Monitor to Zephyr on Open Hardware
14mKeynote: Leveraging Zephyr and ML to Bring Smart Devices to Market, Faster
11mLightning Talk: Accelerating Experimentation...
16mLightning Talk: Efficient Inference at the Edge: Performance You Need at the Lowest...
01h 19mMini-conference: RISC-V Support in Zephyr