Welcome to solt’s documentation!

solt is a fast data augmentation library, supporting arbitrary amount of images, segmentation masks, keypoints and data labels. It has OpenCV in its back-end, thus it works very fast.

Features

• Geometric transformations fusion to speed up a series of arbitrary transformations.
• Logging of all transformations for reproducibility.
• Support of any Computer Vision task.
• Easy extendability, interfacing with other libraries.
• High parametrization.
• Nearly 100% code coverage.

How to install

The most recent stable version is available in pip:

pip install solt

You can also fetch the most fresh changes from our GitHub repository:

pip install -U git+https://github.com/MIPT-Oulu/solt

Modules:

• solt
  • solt.core
  • solt.transforms
  • solt.constants
  • solt.utils module

Examples:

• Basics of SOLT
• Using SOLT to augment the keypoints (landmarks)
• Using SOLT for instance and semantic segmentation
• Augmenting Medical Images: Chest X-ray 14 dataset