kritiksoman / GIMP-ML

AI for GNU Image Manipulation Program
https://kritiksoman.github.io/GIMP-ML-Docs/index.html
MIT License
1.39k stars 127 forks source link

Use Python 3 for running models, add Windows support #13

Closed valgur closed 3 years ago

valgur commented 4 years ago

As suggested by @joemarshall in #4, all PyTorch-specific code could be moved to separate files and be run in a Python 3 subprocess. This sounded doable and interesting, so I gave it a shot.

This PR is based on #12 and adds the following changes:

joemarshall commented 4 years ago

That is brilliant, I was just opening up my text editor to do this then I saw the notification! Thanks for that.

joemarshall commented 4 years ago

Hey, I found with bigger images in deeplabv3 it was for some reason running out of memory in the rpc message building, deep inside xmlrpc for the result. zlib compressing it before building the message fixed it.

I can't work out if there's a way to send a pull request to a pull request, but all I changed was import zlib and added zlib compression before sticking it in the binary. e.g. x = [zlib.compress(x.astype(np.uint8).tobytes()), x.shape]

    return ImgArray(zlib.decompress(x[0].data), x[1])
kritiksoman commented 3 years ago

Sorry we were not accepting PR as GIMP was transitioning to python3 and python2 was already deprecated. If you are interested, please contribute to this branch which is meant for python 3.

joemarshall commented 3 years ago

On 19 Jun 2021 05:15, kritiksoman @.***> wrote:

Sorry we were not accepting PR as GIMP was transitioning to python3 and python2 was already deprecated. If you are interested, please contribute to this branchhttps://github.com/kritiksoman/GIMP-ML/tree/GIMP3-ML which is meant for python 3.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/kritiksoman/GIMP-ML/pull/13#issuecomment-864352525, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAK6Y67HEKWJ6OFKOIJJ6CDTTQKUZANCNFSM4NXLGT4Q.

This message and any attachment are intended solely for the addressee and may contain confidential information. If you have received this message in error, please contact the sender and delete the email and attachment.

Any views or opinions expressed by the author of this email do not necessarily reflect the views of the University of Nottingham. Email communications with the University of Nottingham may be monitored where permitted by law.