Building a ML project for a Samsung Galaxy 1/3 Watch

My goal is to build an inference model just like the continuous motion recognition model demonstrated on the Edge Impulse website but for a standard Samsung Galaxy Watch. My background is in data science/Python/ML so the model building step in the middle mostly makes sense but the embedded and EdgeML area is very new to me. First I would like to know if this is a feasible beginner project so would appreciate some advice from the forum!

Here are a couple of questions I feel I need to answer but there are probably questions I haven’t asked because I don’t know what I don’t yet know so by all means feel free to answer questions I should have asked as well.

  1. How to get the data for training from the watch? I see how I can use my mobile phone to collect and train a model but how do I collect motion and other data sensed by the watch (maybe via the phone) to do the same thing.
  2. How to deploy a trained model back to the Galaxy Samsung watch for inference?

Any help will be much appreciated!

Hi Mattr,

Do you know what is the OS that is used on the Samsung Galaxy watch?


Hi @OmarShrit
Here are the details from the watch:
Tizen: 5.5.02
Software version: R805FXXU1HUK2
Knox: 2.5.0

I think @ijdoc has more information than me.

1 Like