Research on Ceramic Tile Defect Detection Based on Enhanced YOLOv5

Traditional ceramic tile defect detection methods are often limited by the finite nature of feature representation and the influence of complex backgrounds, resulting in low model accuracy and generalization ability. This paper proposes a Shuffle Attention Mechanism-based YOLOv5 detection algorithm, aiming to enhance the accuracy and robustness of ceramic tile defect detection. It is compared and analyzed against the YOLOv5 algorithm, as well as improved YOLOv5 algorithms based on three different attention mechanisms: SimAM, GAM, and NAM. Experimental results demonstrate that this algorithm effectively captures crucial features related to defects in ceramic tile images, enabling more accurate detection of various types of ceramic tile defects and reducing false positives. In summary, this study demonstrates the effectiveness and potential of the Shuffle Attention YOLOv5 algorithm in ceramic tile defect detection.