open-edge-platform/anomalib

An anomaly detection library comprising state-of-the-art algorithms and features such as experiment management, hyper-parameter optimization, and edge inference.

What it solves

Anomalib provides a unified framework for benchmarking, developing, and deploying deep learning anomaly detection algorithms. It specifically addresses the difficulty of implementing and comparing state-of-the-art anomaly detection models, particularly for visual data (images and videos), by offering a large collection of ready-to-use algorithms and benchmark datasets.

How it works

Built on PyTorch Lightning, the library provides a modular API and CLI that abstracts away boilerplate code, allowing users to focus on model implementation. It supports a wide range of hardware backends (CPU, NVIDIA CUDA, AMD ROCm, and Intel XPU) and integrates with OpenVINO for accelerated inference on Intel hardware. The platform includes tools for training, inference, hyperparameter optimization (via Weights & Biases and Comet.ml), and experiment tracking.

Who it’s for

It is designed for researchers and developers working on visual anomaly detection, particularly those in industrial settings where detecting defects or irregularities in images or videos is critical.

Highlights

  • Extensive Algorithm Library: The largest public collection of ready-to-use deep learning anomaly detection algorithms.
  • Hardware Flexibility: Native support for multiple accelerators including Intel GPUs and OpenVINO optimization.
  • Low-Code Option: Includes "Anomalib Studio," a web application for training and deploying models with minimal coding, supporting inputs from USB/IP cameras.
  • Deployment Ready: Provides dedicated inference tools and export options for production environments.

Related

  • Project
    openvinotoolkit/openvino

    An open-source toolkit for optimizing and deploying deep learning models across CPU, GPU, and NPU hardware to boost inference performance.

  • Project
    CVHub520/X-AnyLabeling

    A unified cross-platform desktop application for AI-assisted annotation of text, image, video, and multimodal data, integrating a wide range of deep learning models to automate labeling.

  • Project
    roboflow/supervision

    A computer vision toolkit that provides model-agnostic building blocks for data loading, visualization, and dataset management to accelerate the development of vision applications.

  • Project
    Lightning-AI/pytorch-lightning

    A deep learning framework that abstracts away the engineering boilerplate of PyTorch, enabling researchers to pretrain and finetune AI models at scale across various hardware.

  • Project
    uxlfoundation/oneDNN

    A cross-platform performance library providing optimized building blocks for deep learning applications to improve performance across various CPUs and GPUs.