Hello and thank you for your work,
I have a question regarding the post processing for the segmentation output.
In your paper you write: "we use a function to find contours of segmentation output".
What is done in cases where multiple contours are found for a given segmentation output?
Is this used as a flag to disregard the output as faulty, is the contour with greatest area always taken for further processing, or are all the contours combined in some manner?