Noise Fusion-based Distillation Learning for Anomaly Detection in Complex Industrial Environments

2025-06-19

AI Summary

To address poor robustness in industrial defect detection under complex scenarios involving viewpoint, pose, and illumination variations, this paper proposes Heterogeneous Teacher Network (HetNet) with collaborative distillation. Methodologically, it innovatively integrates adaptive local-global feature modeling with a local multi-variate Gaussian noise generation mechanism, establishing an unsupervised anomaly scoring framework that significantly enhances generalization against diverse perturbations. On the MSC-AD benchmark, HetNet achieves an average performance improvement of approximately 10% across all metrics, setting a new state-of-the-art (SOTA). Real-world deployment confirms seamless integration into production lines, enabling robust and real-time defect localization and detection. The core contributions are: (i) the first introduction of a heterogeneous teacher collaborative distillation paradigm; and (ii) dual decoupled enhancement—both at the feature-representation level and in noise modeling—to improve anomaly discrimination and robustness.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Anomaly detection and localization in automated industrial manufacturing can significantly enhance production efficiency and product quality. Existing methods are capable of detecting surface defects in pre-defined or controlled imaging environments. However, accurately detecting workpiece defects in complex and unstructured industrial environments with varying views, poses and illumination remains challenging. We propose a novel anomaly detection and localization method specifically designed to handle inputs with perturbative patterns. Our approach introduces a new framework based on a collaborative distillation heterogeneous teacher network (HetNet), an adaptive local-global feature fusion module, and a local multivariate Gaussian noise generation module. HetNet can learn to model the complex feature distribution of normal patterns using limited information about local disruptive changes. We conducted extensive experiments on mainstream benchmarks. HetNet demonstrates superior performance with approximately 10% improvement across all evaluation metrics on MSC-AD under industrial conditions, while achieving state-of-the-art results on other datasets, validating its resilience to environmental fluctuations and its capability to enhance the reliability of industrial anomaly detection systems across diverse scenarios. Tests in real-world environments further confirm that HetNet can be effectively integrated into production lines to achieve robust and real-time anomaly detection. Codes, images and videos are published on the project website at: https://zihuatanejoyu.github.io/HetNet/

Problem

• Detect anomalies in complex industrial environments with varying conditions
• Improve defect detection accuracy in unstructured manufacturing settings
• Enhance reliability of real-time anomaly detection systems

Innovation

• Collaborative distillation heterogeneous teacher network (HetNet)
• Adaptive local-global feature fusion module
• Local multivariate Gaussian noise generation module