Item type:Conference Paper,

How Similar is Too Different? A DINOv2-Based Defect Detection Method

Loading...
Thumbnail Image

Fulltext URI

Document type

Text/Conference Paper

Additional Information

Date

relationships.isAuthorOf

Flores Jarrin, Juan F.
Sturm, Fabian
Menz, Benjamin
Bocklet, Tobias

Journal Title

Journal ISSN

Volume Title

Publisher

Gesellschaft für Informatik e.V.

Abstract

Problem-free defect detection with neural networks is still a challenge in flexible industrial environments, especially in contract manufacturing. A training-free defect detection system using DINOv2 is proposed to address these challenges of resource-intensive model training in flexible industrial environments. The method leverages patch-level embeddings, where each patch represents a small image region, and uses cosine similarity to detect and localize anomalies without prior training. Two strategies are compared: Method I (all patches) and Method II (least similar patches). Experiments on MVTec AD, VisA, and KSDD2 show that Method II outperforms Method I, with Area Under the Curve improvements of 6 − 8% and achieved results of 91.3%, 82.8%, and 92.8%, respectively. The study underscores the significance of reference image diversity, object orientation, and patch selection. The approach’s simplicity and adaptability make it a scalable solution for industrial visual inspection, eliminating the need for extensive labeled data and individualized model training.

Description

Flores Jarrin, Juan F.; Sturm, Fabian; Menz, Benjamin; Bocklet, Tobias (2025): How Similar is Too Different? A DINOv2-Based Defect Detection Method. INFORMATIK 2025. DOI: 10.18420/inf2025_116. Bonn: Gesellschaft für Informatik e.V.. PISSN: 2944-7682. pp. 1311. Künstliche Intelligenz im Mittelstand – KI-KMU 2025. Potsdam. 16.–19. September 2025

Keywords

Citation

URI

Endorsement

Review

Supplemented By

Referenced By