Right now the output of object classification is an image with either object probabilities, or an object classification -> pixels are labeled according to the class. In case of touching objects of the same class (that were before extracted with e.g. seeded watershed in the ilastik thresholding applet) it is not possible to tell anymore, that these were separate objects.
I've discussed this with @tischi:
Right now the output of object classification is an image with either object probabilities, or an object classification -> pixels are labeled according to the class. In case of touching objects of the same class (that were before extracted with e.g. seeded watershed in the ilastik thresholding applet) it is not possible to tell anymore, that these were separate objects.
A better output would be
this could then be displayed in @tischi's viewer.