Contributions in any form are welcome.
This is an integration project between tensorflow lite for micro-controllers, micropython, ulab and other tools that will help users experiment within TinyML.
New boards, Documentation Improvements, Adding New examples, etc are all welcome.
Please file issues and new boards can be added to the nightly github actions build. Downloading is supported and this is the easiest way for new users to get started. To just download the firmware for their board and flash it.
Please include a pointer to the board config used to build micropython as that will be the base used for adding the microlite and other modules to it.
For example the ESP32 geneneric board config is located here: https://github.com/micropython/micropython/tree/master/ports/esp32/boards/GENERIC
At the moment we only build some of the boards supported by Micropython. The current minimum requirements are 1 MB Flash Size.
To run microspeech you need to have approximately 50 kb available ram within micropython. To run person detection you need approximately 440 kb memory plus space for the framebuffer. This is most likely only possible on boards supporting PS RAM.
In the future it may be possible to strip down what is included to fit into a smaller flash size but its not working yet. Refer to #53
Please file issues to discuss how to improve documentation. Or area's where its confusing or missing key steps.