基于YOLOv5s的高分辨率遥感影像尾矿库检测方法  

High Resolution Remote Sensing Image Detection Method of Tailings Pond Based on YOLOv5s

作　　者：夏鹏久 赵恒谦[1] 杨姿涵 黄思文 吴沿坤 XIA Pengjiu;ZHAO Hengqian;YANG Zihan;HUANG Siwen;WU Yankun(College of Geoscience and Surveying Engineering,China University of Mining and Technology(Beijing),Beijing 100083,China)

机构地区：[1]中国矿业大学(北京)地球科学与测绘工程学院,北京100083

出　　处：《信息与电脑》2022年第12期172-174,共3页Information & Computer

基　　金：北京市大学生科学研究与创业行动计划项目资助(项目编号:C20190244)。

摘　　要：针对尾矿库实地调研困难,费时费力且效果一般的问题,提出了一种基于YOLOv5s模型的尾矿库高分辨率遥感影像检测方法。YOLOv5s模型速度小,检测速度快,对于大型目标具有良好的检测效果。经测试实验结果表明,YOLOv5s模型的提取精度评价指标精确度(Precision,P)、召回率(Recall,R)和平均精度值(Mean Average Precision,mAP)分别达到了0.84,0.678,0.752,能够满足大范围遥感影像尾矿库检测的应用需求。In view of the difficulties in field investigation of tailings ponds and the large amount of manpower, material and financial resources, a high resolution remote sensing image detection method based on YOLOv5s was proposed. The YOLOv5s model has small volume, fast detection speed, and good detection effect for large targets. The results showed that the Precision, Recall and mAP of YOLOv5s model reach 0.84, 0.678 and 0.752, respectively, which can meet the application requirements of tailings pond detection in a wide range of remote sensing images.

关 键 词：YOLOv5s 目标检测 尾矿库 高分辨率遥感 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]