Machine Learning on Microcontrollers; Tiny, Useful and Fun
summary
TinyML systems run inference on-device at under 1 W and often with under 1 MB RAM and a typical bill of materials under $10; emlearn exports scikit‑learn and TensorFlow/Keras models to portable C99 code or MicroPython mpy modules, generating a C header for firmware and supporting small MLPs, CNNs and tree models. Real deployments include accelerometer cow‑behavior classifiers, an in‑ear audio+IMU breathing monitor, an automated toothbrush timer, and a research prototype with a total BOM below $1 that validates feasibility.
AI-written summary · about
chapters (9)
- 0:00Tiny ML introduction
- 1:23Why machine learning on microcontrollers
- 5:21What are microcontrollers and where they are used
- 6:46EML learn project
- 9:16Real-world EML learn
- 12:28Deploying with EML learn C
- 14:15EML learn with MicroPython
- 18:11Summary and resources
- 19:17Q&A
related talks
01h 53mTinyML: Machine Learning’s Future is Tiny and Bright
27mWhat’s TinyML good for
31mKeynote: The next big thing in machine learning is tiny
29mMicrocontrollers + Machine Learning with MicroPython in 1-2-3
32mNext level Machine Learning with TinyML and Python
44mMachine Learning on Microcontrollers With Zephyr and Emlearn
19mWhy Machine Learning on Microcontrollers?
01h 01mML and the IoT: Living on the Edge