Toward Long-Tailed Online Anomaly Detection through Class-Agnostic Concepts

2025-07-22

AI Summary

This paper formally defines and addresses the novel task of Long-Tailed Online Anomaly Detection (LTOAD)—an unsupervised, label-free online setting where neither anomaly samples nor category labels are available, and class distributions follow a long-tailed pattern. We propose a class-agnostic concept learning framework that eliminates reliance on predefined class counts, ground-truth labels, or anomalous exemplars. Our approach integrates self-supervised feature learning, image-level contrastive learning, momentum-based parameter updates, and an online memory bank to enable continual representation refinement. On the MVTec AD benchmark, our method achieves a 4.63% absolute improvement in image-level AUROC over prior state-of-the-art methods—even surpassing some label-dependent approaches—and maintains a 0.53% gain under long-tailed online evaluation. Furthermore, we introduce the first dedicated LTOAD benchmark, facilitating scalable, real-time defect localization in industrial and medical imaging applications.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Anomaly detection (AD) identifies the defect regions of a given image. Recent works have studied AD, focusing on learning AD without abnormal images, with long-tailed distributed training data, and using a unified model for all classes. In addition, online AD learning has also been explored. In this work, we expand in both directions to a realistic setting by considering the novel task of long-tailed online AD (LTOAD). We first identified that the offline state-of-the-art LTAD methods cannot be directly applied to the online setting. Specifically, LTAD is class-aware, requiring class labels that are not available in the online setting. To address this challenge, we propose a class-agnostic framework for LTAD and then adapt it to our online learning setting. Our method outperforms the SOTA baselines in most offline LTAD settings, including both the industrial manufacturing and the medical domain. In particular, we observe +4.63% image-AUROC on MVTec even compared to methods that have access to class labels and the number of classes. In the most challenging long-tailed online setting, we achieve +0.53% image-AUROC compared to baselines. Our LTOAD benchmark is released here: https://doi.org/10.5281/zenodo.16283852 .

Problem

• Addresses long-tailed online anomaly detection without class labels
• Improves anomaly detection in industrial and medical domains
• Outperforms offline methods in online settings

Innovation

• Class-agnostic framework for anomaly detection
• Adaptation to online learning setting
• Outperforms SOTA in offline LTAD