Mastering TinyML

Collection of TinyML examples for popular MCU boards on popular frameworks (repositories).

Tiny Machine Learning (TinyML) is a subdomain of machine learning (ML) for low-end (less than 1 USD) and low-power (mW range) processors (MCU, DSP) constrainted by CPU, SRAM and FlashRAM and running on batteries. Applications are predictive maintenance, wake word, behavior detection ... More details on TinyML.org.

Coming soon ...

Boards

• Arduino Nicla Sense ME
• Arduino Nicla Vision
• Espressif ESP32 Eye
• Espressif ESP32 Cam
• M5Stack ESP32 PSRAM Timer Camera X (OV3660)
• ST B-L475E-IOT01A
• ST B-U585I-IOT02A
• ST STM32F413H-Discovery
• ST STM32F769I-Discovery
• Raspberry Pi RP2340
• Raspberry Pi RP2040
• Greenwaves GapPOC-A
• Greenwaves GapPOC-B
• Seeedstudio Sipeed MAix Bit
• Seeedstudio Wio Lite AI
• Seeedstudio Wio Terminal
• Nucleo STM32N6 (coming soon)

And for power consumption measurement

• Nordic Power Profiler Kit II
• ST X-NUCLEO-LPM01A

MCU Boards are available at FabMaSTIC fablab.

TinyML Frameworks

• TensorFlow Lite Micro (Edge Impulse, Imagimob)
• MindSpore Lite
• ST X-CUBE-AI (for STM32 MCU)
• ST NanoEdgeAIStudio (for STM32 MCU)

Funding

This work is partially funded by Persyval Lab, MIAI and LIG Lab.