Are these going to be packaged into functions?

[muschellij2]

This seems like a great resource for ANTsR, but will be pretty hard to use for most users. Is there a plan to package the models and the processing (image to patches/slices and reorientation) into functions and then distribute them or just give these as examples as such?

I think adding the functionality to pass in an image or image filename and get back and image/filename of the output would be a great addition.

[ntustison]

Admittedly, this isn't an optimal packaging of the examples. I'm still thinking about it. But, just to be clear, are you aware of the main package ANTsRNet?

[stnava]

yes - i think that parts will move to antsrnet as they mature - this is intended for somewhat raw development

[muschellij2]

Yes - I have seen that and that's how I got to these. I'm just wondering how we could package the trained h5 models into a package effectively (and lightweight).

[stnava]

we are looking at git-annex (or via datalad) as a possibility. but this is still in the nascent stage. brian

…

On Thu, Sep 6, 2018 at 11:10 AM John Muschelli ***@***.***> wrote: Yes - I have seen that and that's how I got to these. I'm just wondering how we could package the trained h5 models into a package effectively (and lightweight). — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AATyflzC7567gFCYyPMxQEz6eTfcPuaMks5uYTr0gaJpZM4WdHra> .