KairosAD: A SAM-Based Model for Industrial Anomaly Detection on Embedded Devices

Uzair Khan, Franco Fummi, Luigi Capogrosso
Department of Engineering for Innovation Medicine, University of Verona, Italy
Paper arXiv Code
🎉 Accepted @ ICIAP 2025 🎉
DIAG abstract
Comparisons of different anomaly detection methods in terms of I-AUROC (vertical axis), inference time (horizontal axis), and the ratios of parameter numbers (circle radius), on the MVTec-AD dataset. Our KairosAD achieves competitive I-AUROC results, while having 78% less parameters with respect to SimpleNet, i.e., the model that achieves the highest AUROC. Instead, with respect to STLM, which represents the state-of-the-art of efficient deep learning method for anomaly detection, KairosAD has 35% of fewer parameters, achieves 4x speedup performance in terms of inference time, with an increase in I-AUROC performance as well.

Abstract

In the era of intelligent manufacturing, anomaly detection has become essential for maintaining quality control on modern production lines. However, while many existing models show promising performance, they are often too large, computationally demanding, and impractical to deploy on resource-constrained embedded devices that can be easily installed on the product lines of Small and Medium Enterprises (SMEs). To bridge this gap, we present KairosAD, a novel supervised approach that uses the power of the Mobile Segment Anything Model (MobileSAM) for image-based anomaly detection. KairosAD has been evaluated on the two well-known industrial anomaly detection datasets, i.e., MVTec-AD and ViSA. The results show that KairosAD requires 78% fewer parameters and boasts a 4x faster inference time compared to the leading state-of-the-art model, while maintaining comparable AUROC performance. We deployed KairosAD on two embedded devices, the NVIDIA Jetson NX, and NVIDIA Jetson AGX. Finally, KairosAD was successfully installed and tested on the real production line of the Industrial Computer Engineering (ICE) Laboratory at the University of Verona.

BibTeX

TBA.