Number of training samples in SNAP

Dear all

I have a question regarding Maximum likelihood classification (maybe same for other classification method as well).

I have some 5 classes and per any classes, I choose 25000 pixels as samples; so based on this I should have 125000 pixels for training. When I put this amount in table 1 (number of training samples) in SNAP, then the run time takes a lot and my computer does not work.


Table1.

So I have to chose small amount of samples. I choose 10000 and SNAP worked with it but I do not know then, how SNAP chose these 10000 sample pixels between 125000 sample pixels.

Is this way correct or not?

Does SNAP take them randomly?

Thanks.

The answer of this question is

  1. The number of samples is the number of pixels within your training polygons which are randomly selected for training. This means, each trained class should have clearly more than 5000 pixels, otherwise you always use the same available pixels.

Source of the post

Accordingly I think there is no need to increase the number of samples, because this number is enough to include most of the differences of you training area polygons.

Yes, you are right . Thanks falah :grinning: