Hello @mkcor! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file doc/examples/applications/plot_human_mitosis.py:

Line 29:80: E501 line too long (137 > 79 characters)

Comment last updated at 2020-06-14 20:11:40 UTC

I'm not a biologist, but the example looks really cool :) After we make our decision on the image hosting, this is ready for me.

CI arror:

/home/travis/build/scikit-image/scikit-image/doc/source/auto_examples/applications/plot_human_mitosis.rst:77:undefined label: sphx-glr-auto-examples-segmentation-plot-multiotsu-py (if the link has no caption the label must precede a section header)```

@sciunto thanks for pointing to the specific error in the CI build; I just looked up the Sphinx docs
https://www.sphinx-doc.org/en/1.7/markup/inline.html#cross-referencing-arbitrary-locations and:

• Doesn't our use case (I mean, throughout ./doc/examples/) correspond to :doc: rather than :ref:?
• I retrieved the label reference from the URLs (https://scikit-image.org/docs/stable/auto_examples/segmentation/plot_watershed.html#sphx-glr-auto-examples-segmentation-plot-watershed-py and https://scikit-image.org/docs/stable/auto_examples/segmentation/plot_multiotsu.html#sphx-glr-auto-examples-segmentation-plot-multiotsu-py, respectively);
• only now did I notice that the existing examples used a different syntax, with underscores and dot (matched in cec5575);
• I wonder why the warning went off with sphx-glr-auto-examples-segmentation-plot-multiotsu-py but not with sphx-glr-auto-examples-segmentation-plot-watershed-py: I hope it's not because the watershed example contains a (unrelated) reference label (i.e., Wikipedia)...

I think one of the figures could use plt.contour on the original image to visualize the result, it's a another way to visualize a binary array which is often useful.

Thank you very much @mkcor for this very nice tutorial! You're right that it'd be great to have a review of the approach and the text by someone from life or medical sciences. Pinging @ThomasWalter and @JDWarner, and also the friends from CellProfiler who are kind enough to let us use their image @0x00b1 @AnneCarpenter

It's certainly fine with the CellProfiler team to make a copy of that image if you deem that preferable.

This should be fixed in tifffile-2020.5.30, which uses a pure Python PackBits decoder in case the imagecodecs package is not installed.

Awesome thanks a lot @cgohlke. @mkcor you can bump the version of tifffile in requirements/default.txt to see if Travis gets completely green.

For me, the label2rgb default colours are too bright and there's not enough of them! We should probably change that but that should be done in another PR. =)

@jni ok, happy to leave the improvement of these colours to subsequent iterations!

I just wanted to acknowledge @alexdesiqueira's suggestion

what if we had a white background?

explicitly: I guess that, if the background is fully transparent -- as conventionally done, as @jni pointed out -- it may appear either black or white, depending on a convention(?) and/or the overlay situation.

@jni:

For me, the label2rgb default colours are too bright and there's not enough of them! We should probably change that but that should be done in another PR. =)

@mkcor please don't mind me then! This PR is great as is right now 🙂

@emmanuelle I was about to update the changelog to acknowledge the reviewers on the biology side; just to make sure, is https://github.com/scikit-image/scikit-image/blob/master/doc/release/release_dev.rst the right file for me to update?

I think one of the figures could use plt.contour on the original image to visualize the result, it's a another way to visualize a binary array which is often useful.

@emmanuelle done 6819653. I thought that the segment counts would be a nice transition into the mitotic index discussion.

PS: By 'binary array' I understand that you mean mono-channel image, right?

@emmanuelle I was about to update the changelog to acknowledge the reviewers on the biology side; just to make sure, is https://github.com/scikit-image/scikit-image/blob/master/doc/release/release_dev.rst the right file for me to update?

Yes exactly! As you're editing this file you can also add an entry about your tutorial (see for example the changelog of 0.17 as an example)

In the gallery the thumbnail is very dark (as in the attached picture)
. Maybe you could take one of the other images of the tutorial like the one with contours which is on a white background? See https://github.com/scikit-image/scikit-image/blob/master/doc/examples/transform/plot_transform_types.py#L129 for how this is done. I tried with the image #2 for example

The CI failure has nothing to do with this PR (I'm working on it BTW).

I think one of the figures could use plt.contour on the original image to visualize the result, it's a another way to visualize a binary array which is often useful.

@emmanuelle done 6819653. I thought that the segment counts would be a nice transition into the mitotic index discussion.

PS: By 'binary array' I understand that you mean mono-channel image, right?

The way you've used contours is cool, indeed. I was thinking more of representing the 0.5 contour of a binary 0-1 / False-True array, so that the contour is the boundary between False and True, it can be a nice viz to superimpose such a contour over the initial image so that you can see both the image and its segmentation. But never mind it's not crucial here.

Thank you for the detailed review, @emmanuelle!
I will load the original image from https://gitlab.com/scikit-image/data/ via pooch in a subsequent PR.
I will also follow up with the scientific reviewers.

thank you all!