Is your feature request related to a problem? Please describe.
something I think Mao mentioned before, which is that other watershed algorithms that use the rgb/grayscale image to inform marker boundaries. OpenCV's method seems to use the RGB image, and maybe would therefore give improved results? Would need to test on use cases common for our group, e.g. leaf and pollen segmentation.
Use annotation coordinates as "seeds" for watershed?
Describe the solution you'd like
Test and implement OpenCV's watershed algorithm into pcv.watershed_segmentation if the outcomes are improved.
Describe alternatives you've considered
Leave it the way it is.
Is your feature request related to a problem? Please describe. something I think Mao mentioned before, which is that other watershed algorithms that use the rgb/grayscale image to inform marker boundaries. OpenCV's method seems to use the RGB image, and maybe would therefore give improved results? Would need to test on use cases common for our group, e.g. leaf and pollen segmentation.
Use annotation coordinates as "seeds" for watershed?
Describe the solution you'd like Test and implement OpenCV's watershed algorithm into
pcv.watershed_segmentationif the outcomes are improved.Describe alternatives you've considered Leave it the way it is.