AI & ML interests

Focused on image augmentation for machine learning, specifically in computer vision tasks like classification, segmentation, and detection. Offers efficient Python tools for data augmentation to improve model performance and generalization

Organization Card
About org cards

image/png

Albumentations

Efficient Image Augmentation for Machine Learning in Python

Albumentations is a fast, flexible image augmentation library designed for machine learning practitioners working on computer vision tasks. Our aim is simple: provide a comprehensive set of tools that can transform any image to augment your datasets, thereby improving model accuracy and robustness.

Features:

  • Wide Range of Augmentations: Supports geometric transforms, color augmentations, flips, rotations, and more, tailored for classification, segmentation, object detection, and working with key points.
  • Easy Integration: Designed to easily fit into any machine learning pipeline.
  • Performance Optimized: Minimizes CPU/GPU load with efficient implementation.
  • Community Driven: Open to contributions and feedback. We evolve with your needs.
from albumentations import (
    HorizontalFlip, Affine, CLAHE, RandomRotate90,
    Transpose, ShiftScaleRotate, Blur, OpticalDistortion, GridDistortion, HueSaturationValue,
    GaussNoise, MotionBlur, MedianBlur,
    RandomBrightnessContrast, Flip, OneOf, Compose
)
import numpy as np

def strong_aug(p=0.5):
    return Compose([
        RandomRotate90(),
        Flip(),
        Transpose(),
        GaussNoise(),
        OneOf([
            MotionBlur(p=0.2),
            MedianBlur(blur_limit=3, p=0.1),
            Blur(blur_limit=3, p=0.1),
        ], p=0.2),
        Affine(translate_percent=0.0625, scale=(0.8, 1.2), rotate_limit=(-45, 45), p=0.2),
        OneOf([
            OpticalDistortion(p=0.3),
            GridDistortion(p=0.1)
        ], p=0.2),
        OneOf([
            CLAHE(clip_limit=2),
            RandomBrightnessContrast(),
        ], p=0.3),
        HueSaturationValue(p=0.3),
    ], p=p)

image = np.ones((300, 300, 3), dtype=np.uint8)
mask = np.ones((300, 300), dtype=np.uint8)
whatever_data = "my name"
augmentation = strong_aug(p=0.9)
data = {"image": image, "mask": mask, "whatever_data": whatever_data, "additional": "hello"}
augmented = augmentation(**data)
image, mask, whatever_data, additional = augmented["image"], augmented["mask"], augmented["whatever_data"], augmented["additional"]

models

None public yet

datasets

None public yet