Image Pattern Recognition to position

I am currently orienting in some DL environments and was trying to find a point where to start.

My current goal is to feed black images that have small white gaussian points on them and extract the position and see if they adhere to a pattern (triplets, doublets, etc).

Since I could define the pattern as a mathematical entity (2 or more gaussian points a certain distance apart), but also as two ‘images’ a distance apart I am not entirely sure what the best approach is.

Is there a tutorial or some guide that I could use to get started with this?