Domain-independent anomalies datasets (adaptions of the MVTec Anomaly Detection dataset)

2024/06/17 by Jonas Bühler, Jonas Fehrenbach, Bühler, Jonas +6
Medicine · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Respiratory viral infections research

paper · pdf · doi:10.48550/arxiv.2407.02910

openalex publication_date 2024/06/17 · openalex created_date 2024/07/06 · openalex updated_date 2026/07/28

Abstract

A modification of anomalib which adds new approaches derived of PatchCore (Labeled PatchCore and Spatial Embedding MLP) and allows to train and test these models in a domain generalization setting. This is an implementation for the paper Domain-independent detection of known anomalies. If you want to try it and have any questions, feel free to contact us.

Related

• GeneralAD: Anomaly Detection Across Domains by Attending to Distorted Features
• AnomalousPatchCore: Exploring the Use of Anomalous Samples in Industrial Anomaly Detection
• Industrial Anomaly Detection with Domain Shift: A Real-world Dataset and Masked Multi-scale Reconstruction
• ADNet: A Large-Scale and Extensible Multi-Domain Benchmark for Anomaly Detection Across 380 Real-World Categories
• The MVTec AD 2 Dataset: Advanced Scenarios for Unsupervised Anomaly Detection