---
id: albucore
version: "0.2.13"
license: MIT
license_treatment: permissive
maintenance: active
---

# albucore — High-performance image processing functions for deep learning and computer vision.

License: permissive · Maintenance: active · Downloads: 3.9M/mo

## What it is and what it does

Albucore is a library of optimized image processing functions designed as the foundation for AlbumentationsX, an image augmentation library. It implements arithmetic operations (multiply, add, power, add_weighted, multiply_add), elementwise math (exp, log, sqrt), and normalization functions that work on images in channel-last format with explicit channel dimensions. The core feature is automatic backend routing: each operation benchmarks across NumPy, OpenCV, NumKong, StringZilla, and eligible PyTorch implementations and selects the fastest path for the given input dtype, size, layout, and channel count.

Most functions support uint8 and float32 dtypes; elementwise operations (exp, log, sqrt) are float32-only. Images must follow strict shape conventions: single images as (H, W, C), batches as (N, H, W, C), and 3D volumes as (D, H, W, C), with channel dimension always required even for grayscale. The library depends on numpy, numkong, and stringzilla at runtime, and requires Python 3.10+. Full functionality also requires a pre-installed PyTorch build (CPU, CUDA, or MPS selected separately) and OpenCV.

Use it for:

- Accelerate image augmentation pipelines in deep learning workflows by routing arithmetic and normalization operations to the fastest available backend.
- Preprocess uint8 or float32 images for computer vision models with automatic dtype-aware optimization.
- Normalize image batches using precomputed statistics (e.g., ImageNet mean/std) via lookup tables for uint8 or fused NumPy operations for float32.
- Apply elementwise transformations (exponential, logarithm, square root) to large float32 image arrays with automatic selection between NumPy and OpenCV.
- Build custom image processing pipelines that adapt to input characteristics without manual backend selection.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Albucore provides optimized atomic image processing functions that automatically select the fastest implementation (NumPy, OpenCV, NumKong, StringZilla, or PyTorch) based on input characteristics for efficient uint8 and float32 image manipulation.

Yes, with conditions. Albucore is actively maintained, has no known vulnerabilities, and offers genuine performance value through automatic backend routing for image operations. Install it if you are building image augmentation or preprocessing pipelines and want to avoid manual optimization. The main friction is the mandatory separate PyTorch installation and the requirement for Python 3.10+; ensure your environment meets these prerequisites before installing.

## Install

pip install albucore
uv add albucore
poetry add albucore

## Installing albucore

Before you install: Low friction: pure Python wheel with three runtime dependencies (numpy, numkong, stringzilla). Actively maintained with a release 1 day old. Requires Python 3.10+ and a separate PyTorch installation; the package itself installs cleanly but full functionality depends on pre-installing a platform-specific PyTorch build.

License in practice: MIT license (permissive): you can use, modify, and distribute Albucore freely in commercial and private projects with minimal restrictions, provided you include the license notice.

Quickstart:

pip install albucore numpy

import numpy as np
import albucore

image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8)
result = albucore.multiply(image, 1.5)

Requires Python 3.10+. Full functionality requires a separate PyTorch installation (CPU, CUDA, or MPS) before installing Albucore; base install works but current public API requires both OpenCV and PyTorch.

Verify before relying:

- Performance improvement magnitude and conditions under which each backend (NumPy, OpenCV, NumKong, StringZilla, PyTorch) is selected
- Whether all documented functions in the description excerpt are fully implemented and stable in version 0.2.13
- Compatibility and performance characteristics with different PyTorch versions and CUDA/MPS configurations

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 3.9M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

optimized image processing functions, fast image arithmetic operations, automatic backend routing for images, numpy opencv image acceleration, deep learning image preprocessing, image-processing, performance-optimization, deep-learning

[View on SkillFed](https://skillfed.io/packages/albucore) · [View on PyPI](https://pypi.org/project/albucore/)