典型文献
Automatic Detection of Weld Defects in Pressure Vessel X-Ray Image Based on CNN
文献摘要：
The visual automatic detection method based on artifi-cial intelligence has attracted more and more attention.In order to improve the performance of weld nondestructive defect detection,we propose DRepDet(Dilated RepPoints Detector).First,we ana-lyze the weld defect dataset in detail and summarize the distribu-tion characteristics of weld defect data,that is,the defect scale is very different and the aspect ratio distribution range is large.Sec-ond,according to the distribution characteristics of defect data,we design DResBlock module,and introduce dilated convolution with different dilated rates in the process of feature extraction to expand the receptive field and improve the detection performance of large-scale defects.Based on DResBlock and anchor-free detec-tion framework RepPoints,we design DRepDet.Extensive experi-ments show that our proposed detector can detect 7 types of de-fects.When using combined dilated rate convolution network in detection,the AP50 and Reca1150 of big defects are improved by 3.1％and 3.3％respectively,while the performance of small de-fects is not affected,almost the same or slightly improved.The fi-nal performance of the whole network is improved a large margin,with 6％AP50 and 4.2％Reca1150 compared with Cascade R-CNN and 1.4％AP50 and 2.9％Reca1150 compared with Rep-Points.
文献关键词：
中图分类号：
作者姓名：
XIAO Wenkai;FENG Xiang;NAN Shuiyu;ZHANG Linlin
作者机构：
Department of Computer Science and Engineering,East China University of Science and Technology,Shanghai 200237,China;Shanghai Engineering Research Center of Smart Energy,Shanghai 201103,China;Shanghai Aino Industrial Technology CO.,LTD,Shanghai 201612,China
文献出处：
引用格式：
[1]XIAO Wenkai;FENG Xiang;NAN Shuiyu;ZHANG Linlin-.Automatic Detection of Weld Defects in Pressure Vessel X-Ray Image Based on CNN)[J].武汉大学自然科学学报（英文版）,2022(06):489-498
A类：
DRepDet,RepPoints,DResBlock,Reca1150
B类：
Automatic,Detection,Weld,Defects,Pressure,Vessel,Ray,Image,Based,visual,automatic,detection,method,artifi,cial,intelligence,has,attracted,more,attention,In,order,performance,weld,nondestructive,Dilated,Detector,First,ana,lyze,dataset,detail,summarize,characteristics,that,scale,very,different,aspect,ratio,distribution,range,large,Sec,according,design,module,introduce,dilated,convolution,rates,process,feature,extraction,expand,receptive,field,defects,anchor,free,framework,Extensive,experi,ments,show,our,proposed,detector,can,types,When,using,combined,network,AP50,big,improved,by,respectively,while,small,not,affected,almost,same,slightly,nal,whole,margin,compared,Cascade
AB值：
0.469168
相似文献
机标中图分类号，由域田数据科技根据网络公开资料自动分析生成，仅供学习研究参考。
