Albumentations

Added serialization / deserialization

- Now we can define transformations in a python dictionary, `json`, `yaml` files and they will be deserialized and used in the code.
- Now we can define transformations in the code and serialize them in python dictionary, `json` and `yaml` files.

[**Jupyter notebook with an example**](https://github.com/albu/albumentations/blob/master/notebooks/serialization.ipynb)

Special thanks to creafz

Added new transformations

- [Lambda](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Lambda)
- [GaussianBlur](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.GaussianBlur)
- [ChannelDropout](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.ChannelDropout)
- [CoarseDropout](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.CoarseDropout)
- [RandomSnow](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomSnow)
- [RandomRain](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomRain)
- [RandomFog](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomFog)
- [RandomSunFlare](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomSunFlare)
- [RandomShadow](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomShadow)

Special thanks to vfdev-5 ternaus BloodAxe kirillbobyrev

Bugfixes and improvements

- Bugfix in [ToGray](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomShadow)
- Bugfix in [ShiftScaleRotate](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.ShiftScaleRotate)
- Bugfix in [GaussNoise](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.GaussNoise)
- Added `fill_value` parameter to [CutOut](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Cutout)
- SpeedUp in [RandomBrightnessContrast](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomBrightnessContrast)

Special thanks to qubvel ternaus albu BloodAxe

Added support for the keypoint transformations to

* [CenterCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.CenterCrop)
* [Flip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Flip)
* [HorizontalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.HorizontalFlip)
* [IAAAffine](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAAAffine)
* [IAACropAndPad](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAACropAndPad)
* [IAAFliplr](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAAFliplr)
* [IAAFlipud](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAAFlipud)
* [IAAPerspective](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAAPerspective)
* [IAAPiecewiseAffine](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.imgaug.transforms.IAAPiecewiseAffine)
* [PadIfNeeded](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.PadIfNeeded)
* [RandomCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomCrop)
* [RandomRotate90](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomRotate90)
* [RandomScale](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomScale)
* [RandomSizedCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomSizedCrop)
* [Rotate](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Rotate)
* [ShiftScaleRotate](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.ShiftScaleRotate)
* [VerticalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.VerticalFlip)

**[Notebook with an example](https://github.com/albu/albumentations/blob/master/notebooks/example_keypoints.ipynb)**

Special thanks to the Evegene Khvedchenya (BloodAxe) for the work.

Added an option to apply the same transformation to the more than one target of the same type.

The possible use case are image2image or stereo-image pipelines.

**[Notebook with an example](https://github.com/albu/albumentations/blob/master/notebooks/example_multi_target.ipynb)**

Special thanks to Alexander Buslaev (albu) for the work.

Added new transformations
* [RandomCropNearBBox](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomCropNearBBox)
* [RandomSizedBBoxSafeCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomSizedBBoxSafeCrop)
* [SmallestMaxSize](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.SmallestMaxSize)

Speed up in
* [Normalize](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Normalize)
* [Flip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Flip)
* [HorizontalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.HorizontalFlip)
* [VerticalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.VerticalFlip)
* [ElasticTransform](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.ElasticTransform)

Bug fixes
* Fix for Compose with multiprocessing DataLoaders.
* Fix in SmallestMaxSize for multiclass masks.
* Fix in [RandomBrightness](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomBrightness)
* Fix in [RandomContrast](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomContrast)

And many others.

Additional
* Performance benchmark was extended to the [Augmentor](https://github.com/mdbloice/Augmentor) and [Solt](https://github.com/MIPT-Oulu/solt) libraries.
* Added table to Readme that shows all implemented transformations with the set of possible targets: images, bounding boxes, masks, key points. (Special thanks to Alex Parinov creafz )
* The library can be installed in anaconda.

Contributors
BloodAxe albu creafz ternaus erikgaas marcocaccin libfun DBusAI alexobednikov StrikerRUS IlyaOvodov ZFTurbo Vcv85 georgymironov LinaShiryaeva vfdev-5 daisukelab cdicle

Bounding boxes support

Transformations that support bounding boxes

The main change in this release is the addition of the operations on bounding boxes to the
* [Flip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Flip)
* [VerticalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.VerticalFlip)
* [HorizontalFlip](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.HorizontalFlip)
* [Transpose](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Transpose)
* [RandomRotate90](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomRotate90)
* [LongestMaxSize](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.LongestMaxSize)
* [Resize](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Resize)
* [RandomScale](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomScale)
* [Crop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.Crop)
* [RandomCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomCrop)
* [CenterCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.CenterCrop)
* [RandomSizedCrop](https://albumentations.readthedocs.io/en/latest/api/augmentations.html#albumentations.augmentations.transforms.RandomSizedCrop)
* [IAAAffine](https://albumentations.readthedocs.io/en/latest/api/imgaug.html#albumentations.imgaug.transforms.IAAAffine)

Supported formats

Currently supported the following formats for the bounding boxes:
1. COCO: `[x_min, y_min, width, height]`, ex `[97, 12, 150, 200]`
2. Pascal VOC: `[x_min, y_min, x_max, y_max]`, ex `[97, 12, 247, 212]`

Bounding box filtering
It may happen that after the transformation a big part of the bounding box was cropped and it is needed to exclude such boxes.

We support such a bounding box filtering based on the:
* Bounding box area, measured in pixels.
* Visible box area, measured in percent.

Smaller changes
* Added support for 8-bit images.
* We changed all `np.random` occurrences to `random` due to the numpy behavior reported in https://github.com/pytorch/pytorch/issues/5059
* Multiple bugfixes.

Added notebooks with examples
* [How to migrate from torchvision to albumentations.](https://github.com/albu/albumentations/blob/master/notebooks/migrating_from_torchvision_to_albumentations.ipynb)
* [How to apply the transformation to the **classification** problems.](https://github.com/albu/albumentations/blob/master/notebooks/example.ipynb)
* [How to apply transformations to the **detection** problems.](https://github.com/albu/albumentations/blob/master/notebooks/example_bboxes.ipynb)
* [How to apply transformations to the **segmentation** problems.](https://github.com/albu/albumentations/blob/master/notebooks/example_kaggle_salt.ipynb)
* [How to apply transformations to the **non 8-bit** images](https://github.com/albu/albumentations/blob/master/notebooks/example_16_bit_tiff.ipynb)
* [All in one showcase](https://github.com/albu/albumentations/blob/master/notebooks/showcase.ipynb)