Anomaly Detection Utilizing PatchCore for Reimagining Industrial Visual Inspection

Abstract

In industrial manufacturing, ensuring quality control is critical to maintaining high standards and operational efficiency. Manual defect detection, however, is often time-consuming, error-prone, and costly, thereby driving the need for automated solutions. This paper investigates a technique for industrial anomaly detection (IAD) by utilizing the state-of-the-art PatchCore algorithm in conjunction with the widely recognized MVTec AD dataset. The dataset consists of 5354 high-resolution color images representing diverse objects and textures, including defect-free samples for training and numerous anomalous instances for testing. With over seventy distinct defect types—such as scratches, dents, contamination, and other structural irregularities—the dataset presents substantial challenges for conventional visual inspection (VI) techniques. The approach integrates convolutional neural networks (CNNs) with patch-based feature extraction methods, enabling PatchCore to accurately identify and localize even subtle anomalies within complex industrial imagery. Experimental results demonstrate that PatchCore significantly enhances detection accuracy, reduces false positives, and streamlines the overall inspection process. These improvements have important implications for operational productivity and quality assurance in various industrial sectors, paving the way for more reliable and cost-effective manufacturing practices.  

Received: 31 May 2025 | Revised: 26 September 2025 | Accepted: 22 December 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available at https://www.mvtec.com/company/research/datasets/mvtec-ad.

Author Contribution Statement

Shalini Kumari: Conceptualization, Software, analysis, Resources, Data curation, Writing – original draft. Chander Prabha: Methodology, Validation, Investigation, Writing – review & editing, Visualization, Supervision, Project administration.