I have two closed polygons, drawn as connected straight black lines on a white background. I need to classify such images in to three forms

  1. Two separate polygons
  2. One polygon encloses the other
  3. The two polygons overlap each other.

The polygons vary in sizes and location on the image, and the image contains only the polygons and the white background.

Which neural network architecture should I use to solve this problem?


1 Answer 1


To make it easy take small known CNN network like Alex Net, train like : input is image, output is [1,0,0,] separate, [0,1,0], encloses, [0,0,1] overlap. Cause you task is easy i guess that will be fine. But it can be done without ML just analysing image with constant algorithm by where are the points of poligons.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .