Please read example_for_a_image.py to use the augmentator.

Two augmentators are implemented in this project.

"styleaug" - self-style augmentation. (Thanks to https://github.com/philipjackson/style-augmentation)

"mixstyleaug" - mixed style augmentation. (Thanks to https://github.com/clovaai/WCT2)

Example of geodesic distance calculation is presented in geodesic_distance_torch.py

Thanks to https://github.com/thuml/Domain-Adaptation-Regression.

• Solve the bugs for CPU.
• Available for single channel images (gray images).
• Available for 3D patches.
• More optional parameters for mixstyleaug. A random_mix parameter has been added.
• Fix the bugs in example_for_a_image.py for saving images.
• Traditional augmentation.
• AugMix.
• Mixup.
• Copy and paste.
• Add calculation of geodesic distance for datasets estimation.