基于改进Mask⁃RCNN的配电线路语义分割  

Improved Mask⁃RCNN based semantic segmentation for distribution power lines

作　　者：谢世朋[1] 李博 张冬 XIE Shipeng;LI Bo;ZHANG Dong(College of Telecommunications&Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Huai’an Power Supply Branch,State Grid Jiangsu Electric Power Co.Ltd.,Huai’an 223002,China)

机构地区：[1]南京邮电大学通信与信息工程学院,江苏南京210003 [2]国网江苏省电力有限公司淮安供电分公司,江苏淮安223002

出　　处：《南京邮电大学学报（自然科学版）》2021年第6期41-46,共6页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：江苏省重点研发计划(BE2017616)资助项目。

摘　　要：为避免配电线路检修时带来停电等问题,目前不停电作业已成为电力部门对配电线路进行检修的主要手段之一。为保证作业人员在进行不停电作业时的安全性,精准的不停电作业安全预警是非常重要的。其中对配电线路进行语义分割是不停电作业安全预警中的关键技术。文中提出了一种配电线路语义分割法——基于改进的Mask⁃RCNN语义分割模型,对配电线路实现精准的语义分割。通过现场实地实验,该方法具有较好的分割效果。To avoid power outages caused by the overhaul of distribution power lines,live line work has become one of the important means for China’s transmission and distribution line maintenance.To ensure the safety of operators during live line work,an accurate safety early warning system for live line work is very important.Among them,the semantic segmentation of distribution power lines is the key technology in the safety early warning of live line work.An improved semantic segmentation method for distribution power lines based on improved Mask⁃RCNN is proposed for distribution power line segmentation.The results of field tests proves that the method has a high segmentation accuracy.

关 键 词：不停电作业 配电线路 语义分割 Mask⁃RCNN 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]