🐦 Twitter
| 📺 YouTube
| 🌍 mkme.org
Support this project and become a patron on Eric's Patreon.
Website, Forum and store are at http://mkme.org
Chat with Me: Discord
Eric's Fun with various flavors of machine learning on Arduino and other microcontrollers to Raspberry Pi. It is now possible to run ML directly on microcontroller hardware using only kilobytes of memory. The Arduino Nano 33 BLE Cortex M4 is directly sported as well as the RP2040 from Raspberry Pi foundation.
It seems the Arm/Cortex processors support ML
The hardest part for me has been understanding what the different machine learning softwares did and what hardware they are meant for- I have summarized this for you right here in the readme!
Forum Thread HERE - http://mkme.org/forum/viewtopic.php?f=4&t=1137
My Videos HERE: https://www.youtube.com/mkmeorg
Buy Nano 33 on Amazon https://amzn.to/36bWdsy
Most of my Pi Pico notes can be found in my Raspberry Pi Repo here: https://github.com/MKme/raspberrypi
Buy the Raspberry Pi Pico here: https://amzn.to/2LUCtCN
Person detection with the pico https://www.youtube.com/watch?v=a4NOwO1YTOE&ab_channel=AnchorageTechSolutions
- Learning Textbook TinyML https://amzn.to/2YfcBnC
Can use your phone, Arduino or other dev board to do very small size neural net and model. Nice web GUI and FREE usage for dev. Often promoted by Hackster.io with great videos and coverage Supports:
-
Arduino
-
C++ library
-
AND STM32Cube.AI (STM board)
-
Edge Impulse Youtube channel- great videos and ideas : https://www.youtube.com/c/EdgeImpulse/videos
-
How to connect Nano 33 to Edge Impulse: https://www.youtube.com/watch?v=wOkMZUaPLUM&ab_channel=EdgeImpulse
-
Good Vid: https://www.youtube.com/watch?v=pYq6VgXeYwc&ab_channel=Hackster.io
-
Another Good Vid: https://www.youtube.com/watch?v=yUre8L9DK-8&ab_channel=Hackster.io
-
Support for STM info https://www.edgeimpulse.com/blog/machine-learning-for-all-stm32-developers-with-stm32cube-ai-and-edge-impulse
-
Voice recognition instruction https://www.youtube.com/watch?v=vbIg4Up1Ts0&ab_channel=EdgeImpulse
-
Voice recognition with Arduino Nano 33 : https://www.youtube.com/watch?v=fRSVQ4Fkwjc&ab_channel=Digi-Key
-
How to compile a generic Arduino Library in Edge https://youtu.be/fRSVQ4Fkwjc?t=863
This seems to be the basis/compiler for some others possibly. Few working examples in the videos
-
Web: http://tinyml.org
-
Good YouTube channel with full length videos https://www.youtube.com/c/tinyML/videos
-
Intro to TinyML Digikey series good https://www.youtube.com/watch?v=BzzqYNYOcWc&ab_channel=Digi-Key
-
TinyML for health care devices https://www.youtube.com/watch?v=Uyo5-7NHE-s&ab_channel=tinyML
-
Forum https://forums.tinyml.org/t/welcome-to-tinyml-forums/7
-
Interesting Seminar video: https://www.youtube.com/watch?v=f9SNqDejOB0&ab_channel=tinyML
-
Jan 2021 Neat preso on practical examples- REAL DATA https://www.youtube.com/watch?v=-gLawBn9Sn8&ab_channel=Arm
Grandad of machine learning requiring significant hardware for learning and running
-
Will use this on the Pi robot/rover for vision analysis
-
My Robot Builds/Playlist on YouTube https://www.youtube.com/watch?v=i4qu3vGrniQ&list=PLxyM2a_cfnzjRXpokx_aADmTFS94OSxGf&ab_channel=MKmeLab
TFLite is for mobile and IOT devices
-
Magic wand with Nano 33 https://www.youtube.com/watch?v=Lfv3WJnYhX0&ab_channel=AndriYadi
-
Magic Wand detailed project code https://www.hackster.io/andri/ai-powered-magic-wand-ab1c90
Supports:
-
Raspberry Pi Pico
-
Unknown other devices. Unclear what is needed for installs/support on PC
-
TF Lite Micro Git repo for the Raspberry Pi Pico https://github.com/raspberrypi/pico-tflmicro
-
The TensorFlow website has information on training, tutorials, and other resources. https://www.tensorflow.org/lite/microcontrollers
-
The TinyML Book is a guide to using TensorFlow Lite Micro across a variety of different systems. https://tinymlbook.com/
-
Eric bought the TinyML Book https://amzn.to/2YfcBnC
-
TensorFlowLite Micro: Embedded Machine Learning on TinyML Systems has more details on the design and implementation of the framework. https://arxiv.org/pdf/2010.08678.pdf
STM32Cube.Ai can be run alone or better yet- it is included in Edge Impulse above!!! Handy it can be used to compile for STM right from the web interface. It looks like it might only support one board- ST iot discovery kit? Need to look in to this.