基于YOLO网络的无酸除鳞带钢表面缺陷实时检测  

Real-Time Detection of Surface Defects of Acid-Free Descaled Steel Strip Based on YOLO Network

作　　者：宋紫棋 毛桂庭[1] 侯奔[1] SONG Zi-qi;MAO Gui-ting;HOU Ben(State Key Laboratory of Exploitation and Utilization of Deep-Sea Mineral Resources,Changsha Research Institute of Mining and Metallurgy Co Ltd,Changsha 410012,Hunan,China)

机构地区：[1]长沙矿冶研究院有限责任公司深海矿产资源开发利用技术国家重点实验室,湖南长沙410012

出　　处：《矿冶工程》2022年第2期136-140,共5页Mining and Metallurgical Engineering

基　　金：中国五矿集团有限公司科技专项计划项目(2019ZXB03);湖南省自然科学基金面上项目(2021JJ30696)。

摘　　要：为了对无酸除鳞后的热轧钢带表面进行检测,建立了一整套无酸除鳞钢带表面缺陷数据集,然后通过改进的YOLO目标检测算法对缺陷数据集进行训练,得到了DF-YOLO检测模型。实验结果证明,DF-YOLO检测模型可以实现钢带表面缺陷的实时检测。In order to detect the surface defects of hot rolled steel strip after acid-free descaling,data sets of surface defect were established for acid-free descaled steel strip,and then trained by the improved YOLO target detection algorithm.Finally,the obtained DF-YOLO detection model was applied in the detection.The results show that the DF-YOLO detection model can realize the real-time detection of surface defects of steel strip.

关 键 词：钢带表面处理 磨料水射流 无酸除鳞 表面缺陷 YOLO 实时检测 

分 类 号：TG142[一般工业技术—材料科学与工程]