![]() |
This branch is under development. Dedicated for GIMP 3 and Python 3. ⭐ ⭐ ⭐ ⭐ are welcome. Waiting for GIMP 3 to release officially. |
https://kritiksoman.github.io/GIMP-ML-Docs/index.html
[1] Model Ensembling.
[2] Deep learning inference package for different computer vision tasks.
[3] Bridge gap between CV research work and real world data.
[4] Add AI to routine image editing workflows.
Welcome people interested in contribution !!
Join us on Slack -->
Contribution guidelines available --> Link.
import cv2
import gimpml
image = cv2.imread('sampleinput/img.png')
out = gimpml.kmeans(image)
cv2.imwrite('output/tmp-kmeans.jpg', out)
out = gimpml.deblur(image)
cv2.imwrite('output/tmp-deblur.jpg', out)[1] Install GIMP 2.99.6 (Only windows and linux)
[2] Clone this repository: git clone https://github.com/kritiksoman/GIMP-ML.git
[3] Change branch :
git checkout --track origin/GIMP3-ML
[3] On linux, run for GPU/CPU:
bash GIMP-ML/install.bat
On windows, run for CPU:
GIMP-ML\install.bat
On windows, run for GPU:
GIMP-ML\install.bat gpu
[4] Follow steps that are printed in terminal or cmd.
FYI: weights link --> Link
| Windows | Linux |
|---|---|
![]() |
![]() |
| Name | License | Dataset |
|---|---|---|
| deblur | BSD 3-clause | GoPro |
| faceparse | MIT | CelebAMask-HQ |
| coloring | MIT | ImageNet |
| monodepth | MIT | Multiple |
| super-resolution | MIT | ImageNet |
| matting | Non-commercial purposes | Adobe Deep Image Matting |
| semantic-segmentation | MIT | ADE20K |
| kmeans | BSD | - |
| dehazing | MIT | Custom |
| denoising | GPL3 | BSD68 |
| enlighten | BSD | Custom |
| interpolate-frames | MIT | HD |
| inpainting | CC BY-NC 4.0 | CelebA, CelebHQ, Places2, Paris StreetView |
| Detect Objects | Apache-2.0 | COCO |
| Filter Folder | Apache-2.0 | COCO |
| Canny Edge | Apache-2.0 | - |
Please cite using the following bibtex entry:
@article{soman2020GIMPML,
title={GIMP-ML: Python Plugins for using Computer Vision Models in GIMP},
author={Soman, Kritik},
journal={arXiv preprint arXiv:2004.13060},
year={2020}
}
GIMP-ML is , but each of the individual plugins follow the same license as the original model's.






