/unsupervised-anomaly-detection

This repository describes the implementation of an unsupervised anomaly detector using the Anomalib library.

Primary LanguageJupyter NotebookMIT LicenseMIT

made-with-python MIT Licence Ask Me Anything ! Open Source Love Tweet

Unsupervised anomaly detection using Anomalib

Description

This repository describes the implementation of an unsupervised anomaly detector on metallic nuts using the Anomalib library. Thereby we evaluate several state-of-the-art deep learning models such as PaDiM, PatchCore, STFPM, FastFlow and Reverse Distillation.

The data used was The MVTEC Anomaly Detection Dataset (MVTec AD), but only the metal nut dataset was used. The training was performed locally on a laptop with an NVIDIA GeForce GTX 1050 Ti GPU and Ubuntu 20.04 LTS operating system.

It is recommended to download the dataset from this link, and organize the dataset in the format shown in the main notebook.

The implementation is fully described in the main notebook: unsupervised-anomaly-detection.ipynb.

figure

Author

Dennis Hernando NÚÑEZ FERNÁNDEZ
https://dennishnf.com

References