Object tracking without binary

Hello everyone,

I am very new to the computer vision toolbox and got a question about object tracking. The examples that I've seen so far all convert the input images into binary by one way or another. Is there a way to avoid this ?

I have a dataset with background value 0 and blobs having their own id (all pixels of an object in a single frame are set to the same value. So blob 5 has all its values set to 5). Since some of these objects are connected it would be good to keep the original segmentation instead of converting the image to binary where the two objects would be detected as a single one.

Therefore I would like to keep the original segmentation and still utilize the tracking. Is this possible? Sorry if this is a dumb question, but I am still figuring out how to properly utilize the toolbox