A Comparative Study of YOLOv8, YOLOv9, and YOLOv10 in Tile Defect Detection

The complex and diverse textures on tile surfaces, along with the variety of defect types, pose significant challenges for automated defect detection. Accurately identifying subtle defects and achieving real-time detection are currently hot research topics. Deep - learning methods represented by YOLO have provided opportunities and solutions to this problem, and YOLOv8, YOLOv9 and YOLOv10 are the three most advanced algorithms in the YOLO series at present, showing excellent detection performance in defect detection. Through an in - depth comparative study of the specific performance of these three algorithms in tile defect detection, this paper provides valuable reference for technology selection and application scenarios in the field of tile defect detection, and better balance between detection efficiency and accuracy.