Component-aware Unsupervised Logical Anomaly Generation for Industrial Anomaly Detection

2025-02-17

AI Summary

In industrial quality inspection, the scarcity of genuine anomalous samples limits supervised anomaly detection, while existing generative methods often produce physically implausible pseudo-anomalies, leading to high false-positive rates. To address this, we propose ComGEN—a component-aware, unsupervised logical anomaly generation framework. ComGEN models part structures via multi-scale component disentanglement and explicitly encodes intrinsic logical constraints among components through text-component-aligned attention-guided residual mapping, enabling interpretable anomaly synthesis without real anomaly samples. It further incorporates iterative reference matching and unsupervised generative editing to enhance generation plausibility and generalizability. Evaluated on MVTecLOCO, ComGEN achieves 91.2% AUROC. Extensive validation on a diesel engine production line and MVTecAD demonstrates significant reductions in false positives and consistent improvements in downstream anomaly detection performance.

Technology Category

• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Anomaly detection is critical in industrial manufacturing for ensuring product quality and improving efficiency in automated processes. The scarcity of anomalous samples limits traditional detection methods, making anomaly generation essential for expanding the data repository. However, recent generative models often produce unrealistic anomalies increasing false positives, or require real-world anomaly samples for training. In this work, we treat anomaly generation as a compositional problem and propose ComGEN, a component-aware and unsupervised framework that addresses the gap in logical anomaly generation. Our method comprises a multi-component learning strategy to disentangle visual components, followed by subsequent generation editing procedures. Disentangled text-to-component pairs, revealing intrinsic logical constraints, conduct attention-guided residual mapping and model training with iteratively matched references across multiple scales. Experiments on the MVTecLOCO dataset confirm the efficacy of ComGEN, achieving the best AUROC score of 91.2%. Additional experiments on the real-world scenario of Diesel Engine and widely-used MVTecAD dataset demonstrate significant performance improvements when integrating simulated anomalies generated by ComGEN into automated production workflows.

Problem

• Unrealistic anomaly generation
• Scarcity of anomalous samples
• Logical anomaly detection in industry

Innovation

• Component-aware unsupervised framework
• Multi-component learning strategy
• Attention-guided residual mapping