Hi,
I would like to detect if there is a black blob in a picture below (school project for detecting defects in fruits):
Right now I just detect if there is a circulat shape (houghcircles) and put 1 if there is and 0 if not - it isn't very accurate. I have to detect the black blobs but some of the blobs are false positive. I have been thinking about teaching an algorithm the good and bad pictures to get the best accuracy. I have only used knn for digit detection would it be a good choice to use it in this exercise? Can you suggest anything other to use?
Thank you!