On-Device Continual Learning for Unsupervised Visual Anomaly Detection in Dynamic Manufacturing

2025-12-15

AI Summary

To address three key challenges in dynamic manufacturing—difficulty in adapting Vision Anomaly Detection (VAD) models, severe edge-resource constraints, and scarcity of labeled data—this paper proposes the first online PatchCore framework supporting unsupervised continual learning on-device. Our method introduces a lightweight *k*-center-based dynamic coreset update mechanism, enabling incremental construction and online optimization of the feature dictionary without cloud-based retraining. It integrates a compact feature extractor with unsupervised pixel-level anomaly scoring, balancing detection accuracy, inference efficiency, and deployment feasibility. Evaluated on an industrial platform, our approach achieves a 12% improvement in AUROC, reduces memory footprint by 80%, and significantly accelerates training compared to batch-wise retraining—thereby fulfilling real-time, low-resource, and self-adaptive quality inspection requirements.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

In modern manufacturing, Visual Anomaly Detection (VAD) is essential for automated inspection and consistent product quality. Yet, increasingly dynamic and flexible production environments introduce key challenges: First, frequent product changes in small-batch and on-demand manufacturing require rapid model updates. Second, legacy edge hardware lacks the resources to train and run large AI models. Finally, both anomalous and normal training data are often scarce, particularly for newly introduced product variations. We investigate on-device continual learning for unsupervised VAD with localization, extending the PatchCore to incorporate online learning for real-world industrial scenarios. The proposed method leverages a lightweight feature extractor and an incremental coreset update mechanism based on k-center selection, enabling rapid, memory-efficient adaptation from limited data while eliminating costly cloud retraining. Evaluations on an industrial use case are conducted using a testbed designed to emulate flexible production with frequent variant changes in a controlled environment. Our method achieves a 12% AUROC improvement over the baseline, an 80% reduction in memory usage, and faster training compared to batch retraining. These results confirm that our method delivers accurate, resource-efficient, and adaptive VAD suitable for dynamic and smart manufacturing.

Problem

• Enables rapid model updates for frequent product changes
• Reduces memory usage on resource-limited edge hardware
• Adapts from limited data without costly cloud retraining

Innovation

• On-device continual learning for unsupervised anomaly detection
• Lightweight feature extractor with incremental coreset updates
• Memory-efficient adaptation eliminating cloud retraining