Recently, I have found out this google teachable machine learning tool
https://youtu.be/T2qQGqZxkD0
Can this tool be applied ,like, in context of tinyml, to solve easy and task-specific tinyml project?
1 Like
I’m not too familiar with this tool, but Microsoft has a similar tool (customvision.ai) where you’d upload images and label them (seems to work in a similar way to the one from Google).
Using the customvision tool, you’d be able to build and extract models in TF or TFLite. You’d have to make the TFLite version optimized/ quantized to make a tinyML model in order to fit a low space/power device (MCU).
Though you should take into account how the MCU would take images and interpret the images. Really depends on the MCU…
1 Like
Very cool! Thank you for sharing.
You can actually go from a Teachable Machine exercise to a TinyML deployment! Encourage folks to try it out.
Has anyone been able to do this? I’ve been trying with my Nano 33 BLE with no success.