Research on surface defect detection method of metallurgical saw blade based on YOLOV5

Abstract

As a typical cutting tool with good performance and high processing efficiency, metallurgical saw blades are widely used in various industries, but surface defects are inevitably generated in the manufacturing process. To solve this problem, this paper proposes a YOLOv5-based surface defect detection model for product quality, which can distinguish three common metallurgical sawblade surface defects with mAP value of 96,1 % in each defect category detection of metallurgical sawblades and detection time of 139,8 ms per image.

Keywords

• metallurgical saw blade
• surface defects
• target detection
• YOLOv5