{"categories":[{"label":"Artificial Intelligence","url":"https://skillfed.io/packages/category/scientific-engineering-artificial-intelligence/8"}],"enrichment":{"capability":"batchgenerators provides data augmentation transforms for 2D and 3D images, including spatial transformations, color adjustments, noise injection, and cropping, designed for medical image analysis and deep learning pipelines.","skillfed_tags":["medical-imaging","data-augmentation","3d-support"],"use_cases":["Augment 3D medical imaging datasets for segmentation tasks while ensuring spatial transforms apply to both images and segmentation masks.","Build training pipelines that apply per-sample augmentation decisions with configurable probabilities rather than batch-wide decisions.","Preprocess multi-modal medical images with controlled misalignments between channels to improve model robustness.","Combine spatial and color augmentations in a composable pipeline and execute them in parallel across multiple worker processes.","Simulate soft-tissue deformations in medical imaging workflows using anatomy-informed augmentation for realistic training data."],"what_it_does":"batchgenerators is a data augmentation framework developed by the Medical Image Computing division at DKFZ for preparing training data in deep learning pipelines. It specializes in augmentations that work uniformly on both 2D and 3D image data\u2014a capability the documentation notes was missing in most competing frameworks. The package provides spatial transforms (mirroring, elastic deformation, rotation, scaling, resampling), color adjustments (brightness, contrast, gamma), noise injection (Gaussian, Rician), and cropping operations, plus specialized augmentations for medical imaging such as anatomy-informed deformations and multi-modal misalignment simulation.\n\nThe typical workflow involves subclassing a DataLoaderBase, implementing a generate_train_batch method to stack transforms using the Compose utility, and then feeding the result into MultiThreadedAugmenter for parallel augmentation across multiple processes. Data flows through the pipeline as a simple Python dictionary with required 'data' and optional 'seg' keys; spatial transforms automatically apply to segmentation masks when present. The package depends on numpy, scipy, scikit-image, scikit-learn, pillow, and pandas for its core operations.","worth_installing":"Yes. batchgenerators is actively maintained, has no known vulnerabilities, and installs with low friction. It is well-suited for medical image analysis and deep learning projects that need 2D/3D augmentation with per-sample control and multithreaded execution. The Apache 2.0 license is permissive. Use it if you need specialized augmentation for medical or scientific imaging."},"id":"batchgenerators","links":{"html":"https://skillfed.io/packages/batchgenerators","md":"https://skillfed.io/packages/batchgenerators.md","pypi":"https://pypi.org/project/batchgenerators/"},"maintenance":{"status":"active"},"meta":{"latest_release":"2026-06-16","license_spdx":null,"license_treatment":"permissive","name":"batchgenerators","python_support":"unspecified","summary":"Data augmentation toolkit"},"popularity":{"monthly_downloads":149364,"position":10994,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"0.25.3"}