Multi-Flow: Multi-View-Enriched Normalizing Flows for Industrial Anomaly Detection

2025-04-04

AI Summary

Existing single-view anomaly detection methods struggle to comprehensively characterize the multi-dimensional attributes of complex industrial products, particularly failing to effectively incorporate multi-view prior knowledge within normalizing flow frameworks. To address this, we propose the first normalizing flow-based multi-view industrial anomaly detection framework. Our method (1) constructs a multi-view graph structure to explicitly model inter-view relationships among product perspectives, and (2) designs a cross-view message-passing mechanism to enable feature alignment and collaborative likelihood estimation. This work pioneers the extension of normalizing flows to the multi-view setting. Evaluated on the real-world multi-view dataset Real-IAD, our approach achieves state-of-the-art performance at both image-level and sample-level anomaly detection, significantly outperforming existing single-view and multi-view baseline methods.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

With more well-performing anomaly detection methods proposed, many of the single-view tasks have been solved to a relatively good degree. However, real-world production scenarios often involve complex industrial products, whose properties may not be fully captured by one single image. While normalizing flow based approaches already work well in single-camera scenarios, they currently do not make use of the priors in multi-view data. We aim to bridge this gap by using these flow-based models as a strong foundation and propose Multi-Flow, a novel multi-view anomaly detection method. Multi-Flow makes use of a novel multi-view architecture, whose exact likelihood estimation is enhanced by fusing information across different views. For this, we propose a new cross-view message-passing scheme, letting information flow between neighboring views. We empirically validate it on the real-world multi-view data set Real-IAD and reach a new state-of-the-art, surpassing current baselines in both image-wise and sample-wise anomaly detection tasks.

Problem

• Bridging gap in multi-view anomaly detection using flow-based models
• Enhancing likelihood estimation by fusing multi-view information
• Improving anomaly detection performance on real-world industrial data

Innovation

• Multi-view normalizing flows for anomaly detection
• Cross-view message-passing enhances likelihood estimation
• State-of-the-art performance on Real-IAD dataset