Is there a method to detect shapes like these accurately and efficiently? I have tried the OpenCv Haar Casacade Classifier which does not work well. These shapes should all be the same class object and can be of different sizes and a little differently shaped (more circular or more angular). In the attached picture there are 4 separate shapes, of which 2 overlap each other. Objects to detect

  • 1
    $\begingroup$ Is the size or area consistent (no really large ones covering 25% of screen, or really small ones only a few pixels across)? These look like outlines of cells from a microscope slide? $\endgroup$ Oct 4, 2021 at 8:15
  • $\begingroup$ @NeilSlater there are langer ones as well but not covering 25% of the screen, but lets say 4 or 5 times as big. These are the smallest ones though. Its an image of a game where I used an Hsv filter to get only what I need to make it easier to detect, these are outlines of the objects $\endgroup$ Oct 4, 2021 at 9:01
  • $\begingroup$ If it is outlines of cells, several researches are done to count "bad" red blood cells from images, and improve the count accuracy by considering the overlapped cells (I think it's for malaria). You might want to look that up for insights. $\endgroup$
    – Sanyou
    Oct 5, 2021 at 5:36


You must log in to answer this question.

Browse other questions tagged .