All Tools
A
OtherFreeOpen Source
AUGMENTOR
Image augmentation pipeline for machine learning training data
MIT
ABOUT
Machine learning models for computer vision often overfit to limited training datasets, producing poor performance on real-world images with different lighting, angles, or noise. Collecting and labeling additional training data is expensive and time-consuming. Augmentor generates synthetic training variations by applying randomized transformations to existing images, effectively expanding dataset diversity and helping models generalize better to unseen inputs without manual data collection.
INSTALL
pip install AugmentorINTEGRATION GUIDE
1. Expand small image classification datasets with rotations, flips, and color jitter
2. Generate diverse training data for object detection models with random crops and zooms
3. Create robust medical imaging models by augmenting X-rays and scans with noise and blur
4. Build balanced datasets for rare classes by oversampling with heavy augmentation pipelines
5. Simulate real-world camera variations for autonomous vehicle perception training
TAGS
image-augmentationcomputer-visiondata-augmentationpythonmachine-learningopen-source