Research Progress on Hot-rolled Strip Surface Defect Detection Based on Machine Vision

Hot-rolled strip is one of the main raw materials in the iron and steel industry, and its surface quality control has always been the key task of intelligent detection in the production process.Aiming at the present situation that automatic online detection of strip surface defects gradually replaced manual detection, the detection methods of strip surface defects were summarized, and the machine-vision-based methods were emphatically introduced, and the application effects of traditional machine vision and deep learning methods of strip surface defects detection were compared and analyzed, and the key technical problems and future development trend of strip surface defect detection were discussed and prospected respectively. The traditional machine-vision-based method of strip defects detection has high detection speed but low accuracy. The mainstream defect detection methods of deep learning have high detection accuracy but slow speed. Therefore, how to improve the accuracy and robustness of the algorithm on the premise of ensuring real-time detection is not only the development trend of detection automation and intelligence, but also the key to deploy in the actual industrial site based on machine vision.