Emmanuelle Gouillart

The code in the setup.py to add compiler flags is basically the same as in #3267, so we'll see which one gets merged first.

@jni hum, this is interesting because with asv I get the same result as you, but when using %timeit I see the improvement ``` In [4]: %timeit b = segmentation.slic(a,...

@jni thanks for checking. Do you think this PR is ok for merging?

I solved the conflicts coming from the merge of #3267, so I think this one is ready @stefanv @soupault

@hmaarrfk good question... I ran a test on a (1500, 2555) RGB image (tiled chelsea :-)) and actually there is a significant slow down with the new version for ``n_jobs...

``` emma@plotty:~/code/scikit-image$ asv dev -b SegmentationSuite · Discovering benchmarks · Running 8 total benchmarks (1 commits * 1 environments * 8 benchmarks) [ 0.00%] ·· Benchmarking existing-py_usr_bin_python3 [ 6.25%] ···...

on master ``` emma@plotty:~/code/scikit-image$ asv dev -b SegmentationSuite · Discovering benchmarks · Running 4 total benchmarks (1 commits * 1 environments * 4 benchmarks) [ 0.00%] ·· Benchmarking existing-py_usr_bin_python3 [...

Thank you @stefanv `dask.delayed` might be worth exploring here, yes. I'm not sure if memoization would be possible thought, I'll think about it.

Hi @jmetz thank you for your PR. What kind of functions would typically be used to compute thresholds in your experience, apart from percentiles? If it's only percentiles, we can...

@yxdragon I also thought adding surface area would be a great addition (see #3812). My idea was also to apply marching cubes on the subimages containing each object. This has...