Beth Cimini
Beth Cimini
Do those look less barfy with the image then being used for watershed, in addition to seed finding?
Cool, much less barfy, just wanted to make sure we didn't have a FIFO issue. Will add a big long comment later but for now just wanted to say thank...
Alright, promised longer comment: - If we're going to offer a "no declumping" option, let's have it behave the same way as IDPrimary, which is not generate seeds at all...
>The speed comparison between regional and local also changes markedly when a distance transform is used - in this scenario regional is slower, which is to be expected since it...
Here's the notebook for distance - you're right that mahotas is quite a bit slower here, but at a footprint of 16, the results are identical, and at a footprint...
>Personally, when playing with the skimage.data.cells3d()[:,1, ...] in the GUI, you can get towards a good segmentation much quicker in 3D using regional than with local (this is using the...
Thanks for summarizing all of this! My thoughts in no particular order - I think, especially since we hope to do more typing of things, that having the two sub-functions...
Minor detail - in general, it would be nice to inherit from the library modules in IDPrimary/Secondary rather than the module-modules (aka, Threshold). I'm guessing for Primary it would be...
I think if it's ok with you, let's keep it open, to remind ourselves about the HTM spec :)
AFAICT, there's no code reason why this should be true, possibly other than the fact that it inherits some stuff from identify. Should be an easy fix.