Release 0.2.1

What's Changed

• fixed the benchmark values by @lext in #55
• Bug in benchmark fixed again by @lext in #56
• Made sure that normalization is always done. by @lext in #57
• bugfix: incorrect serialization (cropping) by @imelekhov in #62
• Can't run unit tests locally on MacOS by @imelekhov in #63
• Added GridMask augmentation 🤔 by @tiulpin in #58
• Gridmask bugfix by @lext in #64
• Fix typo by @tiulpin in #65
• Bump torch from 1.4.0 to 1.13.1 in /benchmark by @dependabot in #83
• Bump black from 22.6.0 to 24.3.0 in /ci by @dependabot in #84
• Bump torch from 1.13.1 to 2.2.0 in /benchmark by @dependabot in #85
• Add step to temporally change project name from setup by @dianapaula19 in #93
• Add production release and build documentation workflow files by @dianapaula19 in #94
• Update color of search navbar and change pip command by @dianapaula19 in #95
• CI updates + codecov integration by @lext in #96
• Update pipelines by @dianapaula19 in #98
• Change version in setup.py temporally for test release by @dianapaula19 in #99
• Update version to 0.2.1 in setup.py by @dianapaula19 in #103

New Contributors

• @imelekhov made their first contribution in #62
• @dependabot made their first contribution in #83
• @dianapaula19 made their first contribution in #93

Full Changelog: v0.1.9...v0.2.1

Improved API, Speed, bugfix a big release

• SOLT is now PyTorch Native
• All the transform subtract the mean and std by default
• Implemented a proper serialization and deserialization
• Allowed to use dict instead of a data container
• Introduced shorter transform names
• Implemented a fair benchmark to compare with other libraries
• Fixed a bug of JPEGCompression (@tiulpin )
• Added IntensityRemappping transform (@soupault )

This release is not backward comparible

JPEG transform and motion blur

A small release that includes JPEG compression augmentation and new blur type - motion blur.

CutOut, KeypointsJitter and tests

This release has some tests-related features and also adds two new transforms.

Detailed description:

• Improved and parametrized many more testes
• Added CutOut data augmentation
• Added KeypointsJitter class that allows to apply random displacements to keypoints

Matrix transformations caching

Matrix transformations are now cached and can be accessed after the application of the transform. New width and height are also stored.

Solely performance improvement release

• Increased the speed of brightness and contrast transforms
• Increased the speed of random crops

Documentation improvements, flip transform extension

v0.1.3

Flip transform fixed and extended. Now allows to perform two-axes fli…

minor bugfixes, CI changes

v0.1.2

pytest version specified

Bugfixes, documentations, CI improvements, docs

v0.1.1

Version fixed.

Release v0.1: Contrast, Brightness, non-distorting rotations

Contrast, brightness and non-distorting rotations have been implemented. Also, one bug in perspective transform, was fixed.