Scaling Edge AI with Elixir and Nerves
summary
Elixir + Nerves run an on-device AI pipeline on a Raspberry Pi with the Hailo HAT, using precompiled AGF models and custom NX‑Halo NIFs to load devices, configure network groups and call infer in real time; an OpenCV binding (EVision) captures frames, YOLOv8 does detection, postprocessing draws boxes and the demo detects people live (two at one point) at under $150 hardware cost and no cloud calls, showing a working, self‑contained edge setup.
AI-written summary · about
chapters (8)
- 0:00Scaling Edge AI with Elixir
- 1:06Discovering Nerves and NX
- 2:55Building NX Halo library
- 5:19Pre/post-processing for YOLO V8
- 6:48Live real-time object detection
- 10:49Benefits of self-contained AI
- 11:53Future of Raspberry Pi clustering
- 13:38Q&A on models and setup
related talks
48mThe Power of Edge AI: Revolutionizing Future of Real-Time, Autonomous Processing
22mEdge AI Use Cases in IoT — Unlocking Intelligence at the Network Edge
10mThe Rise and Impact of Edge AI
38mStreamlining AI Solutions Development: S...
16mRVEdge-Vision: A Fully Open,Ultra-Efficient On-Device AI Platform for Smart Eyewear
50mAI acceleration at the edge
24mAccelerating AI on Edge
36mApplied AI on the edge
