Robust Distribution Alignment for Industrial Anomaly Detection under Distribution Shift

2025-03-19

AI Summary

Industrial anomaly detection suffers from degraded robustness under unknown domain shifts—e.g., illumination variations and sensor drift—due to distributional mismatches between training and deployment environments. To address this, we propose a robust, modality-agnostic detection framework that requires no prior knowledge of the target domain and supports diverse 2D and 3D vision backbones. Our core innovation is the first integration of robust Sinkhorn distance optimization into memory-based feature modeling, enabling end-to-end learning of shift-invariant representations. The method enables rapid adaptation with only a few target-domain samples and unifies adaptation across modalities. Extensive experiments on major industrial benchmarks (MVTec AD, VisA, 3D-AD) demonstrate significant improvements over state-of-the-art methods. Moreover, our approach maintains high recall and low false positive rates under multiple synthetic domain shifts, validating its strong robustness and cross-modal generalization capability.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Anomaly detection plays a crucial role in quality control for industrial applications. However, ensuring robustness under unseen domain shifts such as lighting variations or sensor drift remains a significant challenge. Existing methods attempt to address domain shifts by training generalizable models but often rely on prior knowledge of target distributions and can hardly generalise to backbones designed for other data modalities. To overcome these limitations, we build upon memory-bank-based anomaly detection methods, optimizing a robust Sinkhorn distance on limited target training data to enhance generalization to unseen target domains. We evaluate the effectiveness on both 2D and 3D anomaly detection benchmarks with simulated distribution shifts. Our proposed method demonstrates superior results compared with state-of-the-art anomaly detection and domain adaptation methods.

Problem

• Enhance anomaly detection robustness under domain shifts.
• Optimize Sinkhorn distance for unseen target domains.
• Improve generalization across 2D and 3D anomaly benchmarks.

Innovation

• Memory-bank-based anomaly detection method
• Optimized Sinkhorn distance for robustness
• Enhanced generalization to unseen domains