pip installs another package called gala

[xiuliren]

Hi, thanks for making this useful package.

I used pip install gala, but it installed another package called gala.

https://pypi.org/project/gala/

[jni]

Hi @jingpengw, and thanks for getting in touch! I found that usage of gala had dropped off, and the astronomy community had their own package and asked for the namespace so I gave it to them. Apologies for the inconvenience. I can happily make it available under a different name... Say, gala-segmentation? What do you think?

Which part of gala are you using? I'll point out that unfortunately since many of the dependencies have changed, it's not all working as expected these days, see the discussion in #97

[xiuliren]

Hi @jni, thanks for the explaination.

making it available with gala-segmentation sounds great. I would also like to suggest another name neugala for consideration.

I am only interested in the evaluation part for now. Do you mind if I copy the code to my package chunkflow ? Of course, I'll add some notes that the code is from Gala. I'll maintain the code as well.

[jni]

That makes sense @jingpengw! The license is BSD 3-clause, which allows you to copy things that way, as long as you acknowledge the source and copy the license. I actually recommend you do that, because that's a neat little pure Python chunk and it doesn't make sense to depend on this bigger library just for evaluation.

I'll also point out that the variation of information is now available in scikit-image now, along with adapted rand index, and we would welcome further tweaks/additions if you think you might needs some. Let us know in the issues.

[xiuliren]

thanks! I still prefer to use your code since it has some more options and seems closer to SNEMI3D evaluation.
I have copied the code here

If you think it satisfy the license, we can close this issue?

[jni]

I'll leave it open as a reminder to put this somewhere. =)

I'm quite curious about what features you think are most useful, as it might be worthwhile to port them to scikit-image.

[xiuliren]

Hi Juan, sounds good to me. The split_vi function to get splitter and merger vi. I am also considering adding split_ri to get the splitter and merger rand index.

…

On Thu, Apr 22, 2021 at 9:45 PM Juan Nunez-Iglesias < ***@***.***> wrote: I'll leave it open as a reminder to put this *somewhere*. =) I'm quite curious about what features you think are most useful, as it might be worthwhile to port them to scikit-image. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#99 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2MB5N4MF4CUPCE4A7XL6DTKDGMFANCNFSM42ZQU2GA> .

[jni]

The split_vi function to get splitter and merger vi.
I am also considering adding split_ri to get the splitter and merger rand
index.

Note that both functions in scikit-image return the split values rather than the summed ones. The only thing we don't provide (yet) is the breakdown tables.