Do more with less data
Albumentations is a computer vision tool that boosts the performance of deep neural networks.
The library is widely used in industry, deep learning research, machine learning competitions, and open source projects.
Why Albumentations?
Fast & Performant
Boost model accuracy with highly optimized, benchmark-proven augmentations.
Unmatched Versatility
>100 transforms for images, masks, bounding boxes, keypoints & 3D. Used in medical, satellite, self-driving, and more.
Effortless Integration
Familiar API, similar to torchvision, for easy adoption in PyTorch, TensorFlow, and other frameworks.
Proven & Trusted
Widely adopted in research, competitions (like Kaggle), and commercial applications.
Powerful Features
Versatile Transforms
Pixel-level adjustments (brightness, contrast, noise) and spatial transformations (rotate, scale, flip).
Learn More→Task Agnostic
Consistently handles images, segmentation masks, bounding boxes, and keypoints through any augmentation pipeline.
Learn More→Performance Focused
Highly optimized code ensures minimal overhead, crucial for training large models. See benchmarks.
Learn More→Framework Agnostic
Works seamlessly with PyTorch, TensorFlow, Keras, and other frameworks, using standard NumPy arrays.
Learn More→Extensible
Easily create custom augmentations or pipelines to fit your specific research or application needs.
Learn More→Easy Serialization
Save and load augmentation pipelines using YAML or JSON for reproducibility and sharing.
Learn More→Trusted By Leading Companies
Community Feedback
Community-Driven Project, Supported By
Albumentations thrives on developer contributions. We appreciate our supporters who help sustain the project's infrastructure.
??Community Supporter
??Integration Partner
??Exclusive Partner
Citing
If you find Albumentations useful for your research, please consider citing the paper:
@Article{info11020125,
AUTHOR = {Buslaev, Alexander and Iglovikov, Vladimir I. and Khvedchenya, Eugene and Parinov, Alex and Druzhinin, Mikhail and Kalinin, Alexandr A.},
TITLE = {Albumentations: Fast and Flexible Image Augmentations},
JOURNAL = {Information},
VOLUME = {11},
YEAR = {2020},
NUMBER = {2},
ARTICLE-NUMBER = {125},
URL = {https://www.mdpi.com/2078-2489/11/2/125},
ISSN = {2078-2489},
DOI = {10.3390/info11020125}
}Read the paper: Information, Volume 11, Issue 2