Towards training noise-robust anomaly detection via collaborative adversarial flows

作     者：Cheng, Hao Luo, Jiaxiang Zhang, Xianyong Liu, Haiming Wu, Fan 

作者机构：South China Univ Technol Sch Automat Sci & Engn Guangzhou 510640 Guangdong Peoples R China South China Univ Technol Minist Educ Precis Elect Mfg Equipment Engn Res Ctr Guangzhou 510640 Guangdong Peoples R China South China Univ Technol Key Lab Autonomous Syst & Networked Control Minist Educ Guangzhou 510640 Guangdong Peoples R China Guangdong Polytech Normal Univ Sch Automat Guangzhou 510180 Guangdong Peoples R China Guangdong MOJE Intelligent Equipment CO Ltd Guangzhou 510765 Guangdong Peoples R China 

出 版 物：《MEASUREMENT》 (Meas J Int Meas Confed)

年 卷 期：2025年第242卷

核心收录：

学科分类：08[工学] 080401[工学-精密仪器及机械] 0804[工学-仪器科学与技术] 081102[工学-检测技术与自动化装置] 0811[工学-控制科学与工程] 

基　　金：National Natural Science Foundation of China [62271214, 62173102] Science and Technology Program of Guangzhou, China [2023B01J0037] 

主　　题：Noise-robust anomaly detection Real-world defect inspection Training noise Collaborative adversarial optimization Normalizing flows 

摘      要：Unsupervised visual anomaly detection (UVAD) aims at identifying unobserved anomalies by training a classification model with anomaly-free images, which has been widespread applied to surface inspection of multiple industrial products. However, previous UVAD methods trained by ideal normal samples suffer from performance degradation in real world industry because of the inevitable intrusion of noisy samples into the training set. Addressing this issue, this article introduces a novel strategy, Collaborative Adversarial Flows (CAF), which employs multiple normalizing flows to efficiently filter out noisy samples and accurately construct distributions of normal samples. Unlike previous methods that focused solely on learning accurate representations of normal samples, CAF adds an additional learning objective of inferring noisy samples based on distributional metrics while optimizing representations. Furthermore, CAF establishes a flow-based collaborative adversarial paradigm that promotes high-likelihood samples and rejects low-likelihood noisy samples in optimizing the latent distribution, thereby enhancing the robustness of model under contaminated training set. Comprehensive experiments on the public benchmark MVTec AD under human-made noise intrusion demonstrate that CAF effectively improves the robustness to noisy samples and achieves impressive performance. In electronic manufacturing processes with real industrial noise, the applicability and robustness of CAF have been further proved by the inspection of surface mount devices.