pip install albumentationsx

Fast image augmentation for computer vision

AlbumentationsX is the actively developed library from the Albumentations ecosystem. Build fast, reproducible pipelines for images, masks, bounding boxes, keypoints, and 3D data.

Get startedTry in Explore

Public repository: AGPL-3.0-only Commercial licensing available

Original image and AlbumentationsX transform results

Evidence from the Albumentations ecosystem

Public source-code, research, competition, and artifact evidence show where the Albumentations ecosystem is used. Package dashboards remain separate from adoption evidence.

Evidence updated

Package download dashboards
Live package metrics are tracked separately from the public adoption publication.Albumentations dashboard AlbumentationsX dashboard
Papers citing Albumentations
2,270
Official GitHub repositories
Albumentations AlbumentationsX

Explore by vertical

From microscopy and medical scans to Earth observation, autonomous systems, farms, and factories.

One visual per canonical vertical. Source totals are multi-label and non-additive; illustrative images are not evidence.

Explore adoption evidence

Built for real training pipelines

The targets, performance evidence, and extension points needed for practical computer vision work.

One pipeline, every target

Apply consistent transforms to images, segmentation masks, bounding boxes, keypoints, and 3D data.

View supported targets

Benchmark-backed speed

Evaluate performance through a reproducible public benchmark against other computer vision libraries.

See the benchmark

Fits your stack

Use a NumPy-based interface with custom transforms and serializable pipelines across training frameworks.

Build a custom transform

Used by researchers and engineers

Original public posts from people using Albumentations in scientific imaging, competitions, and vision projects.

  • Testimonial from Claudia Vanea, Research Scientist at Nuance Labs
    “Thank you Vladimir Iglovikov! We used Albumentations to train our digital pathology models extensively during my PhD at Oxford and it's great to see native H&E stain support! Stain-invariance is vital for training models…”

    Claudia Vanea

    Research Scientist at Nuance Labs

    View original post
  • Testimonial from Uladzislau Leketush, ML Engineer
    “One of the key components of my 1st place solution in the Kaggle competition Recod.ai/LUC Scientific Image Forgery Detection was heavy domain-specific augmentation built with Albumentations. For microscopy images, I…”

    Uladzislau Leketush

    ML Engineer

    View original post
  • Testimonial from Christof Henkel, Kaggle Competitions Grandmaster. Top 1 in the world.
    “Great to see that @albumentations is becoming active on twitter. Its such a great library, and so well written. You not only can use the variety of augmentations for computervision but also easily adjust and implement…”

    Christof Henkel

    Kaggle Competitions Grandmaster. Top 1 in the world.

    View original post
View more testimonials

Need commercial terms for a defined deployment?

AlbumentationsX is available under AGPL-3.0-only. The AGPL permits commercial and internal use subject to its terms. Albumentations, LLC also offers separately negotiated commercial terms for organizations that need alternative rights.

View commercial licensing
NumFOCUS Affiliated ProjectAlbumentations is a NumFOCUS Affiliated Project in the open-source scientific computing ecosystem.

Citing Albumentations

If Albumentations supports your research, cite the original 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: Albumentations: Fast and Flexible Image Augmentations