Towards Scalable IoT Deployment for Visual Anomaly Detection via Efficient Compression

2025-05-11

AI Summary

To address the deployment challenges of vision-based anomaly detection (VAD) on resource-constrained industrial IoT edge devices—characterized by limited computational power and bandwidth—this paper presents the first systematic evaluation of lightweight compression strategies (i.e., quantization, low-rank decomposition, and lossy image compression) on representative VAD models (PatchCore and SPADE), analyzing their accuracy–latency trade-offs. We propose an end-to-end adaptation framework that achieves up to 16× data compression on the MVTec AD benchmark, with AUC degradation under 0.5% and inference latency reduced by 60%. Our key contribution is the empirical validation that high-compression-ratio VAD inference incurs negligible performance loss, thereby establishing a scalable, low-overhead paradigm for edge deployment in resource-limited industrial settings.

Technology Category

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

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Virtualization and resource management in Web systems and infrastructures
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Visual Anomaly Detection (VAD) is a key task in industrial settings, where minimizing waste and operational costs is essential. Deploying deep learning models within Internet of Things (IoT) environments introduces specific challenges due to the limited computational power and bandwidth of edge devices. This study investigates how to perform VAD effectively under such constraints by leveraging compact and efficient processing strategies. We evaluate several data compression techniques, examining the trade-off between system latency and detection accuracy. Experiments on the MVTec AD benchmark demonstrate that significant compression can be achieved with minimal loss in anomaly detection performance compared to uncompressed data.

Problem

• Efficient compression for IoT-based visual anomaly detection
• Balancing latency and accuracy in edge device deployments
• Minimizing performance loss in compressed anomaly detection systems

Innovation

• Leveraging compact processing for IoT VAD
• Evaluating compression vs accuracy trade-offs
• Achieving high compression with minimal performance loss