albumentations-team/albumentations ? reverse-engineered prompt

Build me a Python library for image augmentation that I can use in computer vision projects.

I want a simple way to chain together random image edits like crops, flips, color changes, blur, noise, rotation, and resizing, then apply them consistently to images and also to masks, bounding boxes, and keypoints. It should be easy to use for classification, segmentation, and object detection, with one clean API that feels practical in real training code. Please include a few solid example pipelines, clear documentation, and tests so I can trust it works. Make it fast and reliable, and structure it like a real package I could install and use in a deep learning project. If you need to check current best practices for any image processing details, look up the latest docs online.