Value-order Decomposition for Generalist Anomaly Detection

2026-06-12

AI Summary

Industrial anomaly detection suffers from limited generalization to unseen target domains due to data scarcity and the strong coupling of anomaly characteristics with object categories, defect types, and data domains. This work proposes Value-Order Decomposition (VOD), a method that explicitly decouples and suppresses category-, defect-, and domain-specific information—encompassing both real and synthetic anomalies—while preserving separability between normal and anomalous samples and promoting intra-class alignment. VOD is the first approach to systematically bridge three distinct generalization gaps, enabling unified cross-domain, cross-category, and cross-defect-type anomaly detection using only normal samples and synthetic anomalies. Extensive experiments demonstrate that VOD significantly improves generalization performance across multiple industrial and medical benchmarks.

Abstract

Industrial anomaly detection suffers from limited data, making cross-domain generalization particularly challenging. Generalist Anomaly Detection (GAD) aims to train a unified model on a source domain that can effectively detect anomalies in unseen target domains. In the initial semantic feature space, strong entanglement between anomalies and object categories or defect types hinders effective generalization across domains. Recent works address this issue by projecting features into a residual space; however, such methods primarily increase cross-domain overlap for normal features, while anomalous features remain specific to object categories, defect types and data domains, leading to poor alignment and generalization. To address this limitation, we propose Value-order Decomposition (VOD), a simple yet effective technique that bridges \textbf{three types of generalization gaps} across object categories, defect types (including real and synthetic defects), and data domains. VOD disentangles and suppresses object-category-, defect-type-, and domain-specific information, promoting alignment within normal and abnormal samples while preserving their separability, thereby enabling robust generalization across the three gaps. Leveraging the strong alignment between real and synthetic defects within the same object, we perform anomaly detection using only normal and synthetic-abnormal reference, and effectively generalize to unseen real defect types. Experiments on diverse industrial and medical benchmarks demonstrate that our method, using a simple cut-and-paste anomaly simulation strategy, achieves strong generalization across the three gaps.

Problem

• Generalist Anomaly Detection
• cross-domain generalization
• anomaly detection
• domain adaptation
• feature disentanglement

Innovation

• Value-order Decomposition
• Generalist Anomaly Detection
• cross-domain generalization
• feature disentanglement
• synthetic anomaly simulation