GIMP-ML-Hub

Machine Learning plugins for GIMP.

Forked from the original version to improve the user experience in several aspects:

• The PyTorch models are packaged in PyTorch Hub format and are only downloaded as needed. This allows new models to be added more seamlessly, without needing to re-download gigabytes of model weights.
• Models are run with Python 3, saving the needed effort to back-port them to Python 2.
• Fully automatic installation, that has been tested on all major operating systems and distros.
• Errors are now reported directly in the UI, rather than on the command line only.
• Correct handling of alpha channels.
• Automatic conversion between RGB/grayscale as needed by the models.
• Results are always added to the same image instead of creating a new one.
• And many other smaller improvements.

The plugins have been tested with GIMP 2.10 on the following systems:

• macOS Catalina 10.15.5
• Ubuntu 18.04 LTS
• Ubuntu 20.04 LTS (apt-get only, snap is not yet supported)
• Debian 10 (buster)
• Arch Linux
• Windows 10

Installation Steps

1. Install GIMP.
2. Clone this repository: git clone https://github.com/valgur/GIMP-ML-Hub.git
3. On Linux and MacOS run ./install.sh.
4. On Windows:
   • Install Miniconda.
   • Enable execution of Powershell scripts.
   • Run install.ps1.
5. You should now find the GIMP-ML plugins under Layers → GIMP-ML. Feel free to create an issue if they are missing for some reason.

References

MaskGAN

• Source: https://github.com/switchablenorms/CelebAMask-HQ
• Torch Hub fork: https://github.com/valgur/CelebAMask-HQ
• License:
  • CC BY-NC-SA 4.0
  • Copyright (C) 2017 NVIDIA Corporation. All rights reserved.
  • Restricted to non-commercial research and educational purposes
• C.-H. Lee, Z. Liu, L. Wu, and P. Luo, “MaskGAN: Towards Diverse and Interactive Facial Image Manipulation,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019.

Face Parsing

• Source: https://github.com/zllrunning/face-parsing.PyTorch
• Torch Hub fork: https://github.com/valgur/face-parsing.PyTorch
• License: MIT
• Based on BiSeNet:
  • https://github.com/CoinCheung/BiSeNet
  • License: MIT
  • C. Yu, J. Wang, C. Peng, C. Gao, G. Yu, and N. Sang, “BiSeNet: Bilateral segmentation network for real-time semantic segmentation,” in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2018, vol. 11217 LNCS, pp. 334–349.

SRResNet

• Source: https://github.com/twtygqyy/pytorch-SRResNet
• Torch Hub fork: https://github.com/valgur/pytorch-SRResNet
• License: MIT
• C. Ledig et al., “Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 105–114.

DeblurGANv2

• Source: https://github.com/TAMU-VITA/DeblurGANv2
• Torch Hub fork: https://github.com/valgur/DeblurGANv2
• License: BSD 3-clause
• O. Kupyn, T. Martyniuk, J. Wu, and Z. Wang, “DeblurGAN-v2: Deblurring (Orders-of-Magnitude) Faster and Better,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 8877–8886.

MiDaS

• Source: https://github.com/intel-isl/MiDaS
• License: MIT, (c) 2019 Intel ISL (Intel Intelligent Systems Lab)
• R. Ranftl, K. Lasinger, D. Hafner, K. Schindler, and V. Koltun, “Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset Transfer,” 2019.

Monodepth2

• Source: https://github.com/nianticlabs/monodepth2
• Torch Hub fork: https://github.com/valgur/monodepth2
• License:
  • See the license file for terms
  • Copyright © Niantic, Inc. 2019. Patent Pending. All rights reserved.
  • Non-commercial use only
• C. Godard, O. Mac Aodha, M. Firman, and G. Brostow, “Digging Into Self-Supervised Monocular Depth Estimation,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 3827–3837.

Neural Colorization

• Source: https://github.com/zeruniverse/neural-colorization
• Torch Hub fork: https://github.com/valgur/neural-colorization
• License:
  • GNU GPL 3.0 for personal or research use
  • Commercial use prohibited
  • Model weights released under CC BY 4.0
• Based on fast-neural-style:
  • https://github.com/jcjohnson/fast-neural-style
  • License:
    • Free for personal or research use
    • For commercial use please contact the authors
  • J. Johnson, A. Alahi, and L. Fei-Fei, “Perceptual Losses for Real-Time Style Transfer and Super-Resolution,” in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9906 LNCS, 2016, pp. 694–711.

Authors

• Martin Valgur (valgur) – this version
• Kritik Soman (kritiksoman) – original GIMP-ML implementation

License

MIT

Please note that additional license terms apply for each individual model. See the references list for details. Many of the models restrict usage to non-commercial or research purposes only.