I suggest
- clone the current project
- delete all the data
- collect new data with your sensor saving 5s samples
- label the new data
- upload new data and labels
- train and deploy as before
I suggest
I have managed to collect the data my self using my sensor as suggested by you,but the maximum accuracy I can get is like 60% I dont know what the issue could be
@aurel I wanted to add the combination of these feature extraction [‘VAR’, ‘MAV’, ‘WL’, ,‘ZC’,‘SSC’].I have tried to add them in the custom block but i could not clone it . Could you add them instead.
This is our default spectral block so we can’t add those features there. It’s better if you create your own custom block.
Aurelien
@aurel If i added my own custom block with those feature extraction i mentioned, will i be able to see the Feature importance?
I would love to see this implementation. I am trying the same thing but the url gets a 404