Anomalib: A Deep Learning Library for Anomaly Detection

TL;DR

Anomalib is an open-source, modular deep learning library that facilitates the design, implementation, and deployment of unsupervised anomaly detection models with state-of-the-art performance and real-time deployment support.

Contribution

It introduces a comprehensive, reproducible library with plug-and-play algorithms, tools for customization, and deployment features for anomaly detection tasks.

Findings

Achieves top performance on benchmark datasets.

Supports real-time deployment with OpenVINO.

Provides tools for custom algorithm design.

Abstract

This paper introduces anomalib, a novel library for unsupervised anomaly detection and localization. With reproducibility and modularity in mind, this open-source library provides algorithms from the literature and a set of tools to design custom anomaly detection algorithms via a plug-and-play approach. Anomalib comprises state-of-the-art anomaly detection algorithms that achieve top performance on the benchmarks and that can be used off-the-shelf. In addition, the library provides components to design custom algorithms that could be tailored towards specific needs. Additional tools, including experiment trackers, visualizers, and hyper-parameter optimizers, make it simple to design and implement anomaly detection models. The library also supports OpenVINO model optimization and quantization for real-time deployment. Overall, anomalib is an extensive library for the design, implementation, and deployment of unsupervised anomaly detection models from data to the edge.