Lars Grüter
Lars Grüter
Thanks @stefanv! I'm happy with this now. :blush:
> I find it somewhat unintuitive that the priority is given by an image, rather than, say, a dictionary of labels->priority, or an array for which the number at idx...
Sounds good. :blush:
I thought about this some more. There are significant assumptions that the current algorithm can make, that go out of the window if we take labels from the user. First...
> OK, looks like we still need to follow up on https://github.com/scikit-image/scikit-image/pull/4165#issuecomment-1399423776 here Yes, I've been exploring an option that relies solely on sorting by priority to discover all the...
Ready to merge IMO! :tada: I'm somewhat proud of this one now. :blush: https://github.com/scikit-image/scikit-image/pull/4165#issuecomment-1399423776 is addressed and I figured out how to include only objects surface in the KD-tree, while...
With #6714 merged, the table of contents no longer includes signatures (e.g. see [`skimage.draw`](https://scikit-image.org/docs/dev/api/skimage.draw.html)). I disabled signatures there, because in my opinion it didn't really help with readability and much...
See https://github.com/scikit-image/scikit-image/pull/7176 for an attempted fix.
First of all, thank you for the detailed report @tokiAi! I can reproduce the behavior easily. In this case, I agree that the result can be somewhat surprising but I...
Agreed, the documentation could do a better job. > possibly by moving it higher in the list This would require a deprecation cycle. `channel_axis` is the only keyword-only parameter and...