Anomalib is a deep learning library that aims to collect state-of-the-art anomaly detection algorithms for benchmarking on both public and private datasets. Anomalib provides several ready-to-use implementations of anomaly detection algorithms described in the recent literature, as well as a set of tools that facilitate the development and implementation of custom models. The library has a strong focus on image-based anomaly detection, where the goal of the algorithm is to identify anomalous images, or anomalous pixel regions within images in a dataset. Anomalib is constantly updated with new algorithms and training/inference extensions, so keep checking!

Sample Image

Structure of the Documentation

This documentation is divided into the following sections:


The Tutorials section contains all the necessary information regarding setting up and installing the package. It also includes steps needed to train, export, infer models, perform HPO, benchmarking and logging.

Reference Guide


This page contains all the models implemented in the repository as well as their API. It is the developer’s responsibility to update this page when a new model is added to the repo.

API Reference

This page lists all the modules, classes and functions available within the anomalib package. This page is update automatically for the following modules:

  • cli

  • config

  • data

  • model

  • post_processing

  • metrics

  • loggers

  • hpo

  • callbacks

If a change is made to any of these modules, then the document will be automatically updated. However, if a new algorithm is introduced, then it must be added to Algorithms.

Developer Guide

This section contains all the guidelines for those who would like to contribute to the development of anomalib and to know more about the developer tools.

Citing the repository

You can cite this repository as

      title={Anomalib: A Deep Learning Library for Anomaly Detection},
      author={Samet Akcay and
            Dick Ameln and
            Ashwin Vaidya and
            Barath Lakshmanan and
            Nilesh Ahuja and
            Utku Genc},

Indices and tables