Self-Adaptive Gamma Context-Aware SSM-based Model for Metal Defect Detection

The quality of metal surfaces is critical in various industrial applications, including aerospace, manufacturing, and container transportation. Surface defects, such as cracks, dents, and scratches, not only compromise the structural integrity and aesthetics of metal products but also lead to significant economic losses if left undetected. As a result, the accurate and efficient detection of metal surface defects has become an essential task in industrial quality control. In recent years, the adoption of deep learning techniques has significantly advanced the performance of defect detection systems [1]. Convolutional neural networks (CNNs) and transformer-based models have demonstrated exceptional capabilities in handling complex image-based tasks, enabling automated and reliable defect detection. However, several challenges remain: 1) Metal defect often exhibits varied and localized features, making effective multi-scale feature aggregation vital for improving detection accuracy.