ViT-Core+: Lightweight Anomaly Detection Using Cut-Paste and Transformer-Based Feature Extractor

In modern smart manufacturing, real-time anomaly detection is critical, yet state of the art (SOTA) models like PatchCore are often too computationally demanding for practical deployment on resource-constrained edge devices. This creates a crucial gap between algorithmic potential and industrial application. To bridge this gap, we build upon our previous work on ViT-Core and propose ViT-Core +, an enhanced lightweight and efficient model for anomaly detection. Our primary innovation is a refined self-supervised, Cut-Paste-based transfer learning stage that better adapts the efficient Swin Transformer backbone to the target data distribution without requiring complex training. Evaluated on the comprehensive MVTec AD benchmark, ViT-Core+ demonstrates a drastic reduction in computational overhead, with memory usage decreased by 51.3% and inference time by 52.1% compared to the PatchCore baseline, while improving upon the efficiency of the original ViT-Core. This optimization is achieved with a statistically insignificant difference in image-level classification performance. Moreover, ViT-Core+ excels in anomaly localization, achieving a new state-of-the-art pixel-level AUROC of 0.9822. Consequently, ViT-Core+ presents an optimal balance of accuracy and efficiency, providing a practical and scalable solution that enables the widespread deployment of high-performance, real-time quality inspection systems directly on existing industrial hardware.