albumentations-team/albumentations

Albumentations is a Python library for image augmentation. Image augmentation means taking the photos you already have for training a machine learning model and creating new training samples from them by flipping, cropping, changing brightness, and applying other transforms. The README explains that this technique is used in deep learning and computer vision tasks to increase the quality of trained models, because the model effectively sees more varied examples. The library exposes a simple, unified interface that the README says works with the major computer vision tasks: classification (deciding what is in an image), segmentation (marking which pixels belong to which object), object detection (drawing boxes around things), and pose estimation. The same code can also handle masks, bounding boxes, and keypoints alongside the image, so the augmentations stay aligned with the labels. The README lists more than seventy augmentations available out of the box. You would reach for Albumentations when you are training a vision model and want to squeeze more out of a limited dataset, especially in projects using PyTorch or TensorFlow (deep learning frameworks named in the README). One important caveat from the README itself: the original Albumentations is no longer actively maintained, with the last update in June 2025, new development has moved to a successor project called AlbumentationsX, which the authors describe as a drop-in replacement under a different, dual license. The full README is longer than what was provided.