基于K-Means的铁路货运车辆异物识别方法  

A Method for Detecting Foreign Objects on Railway Freight Vehicles Based on K-Means

作　　者：蓝志鹏 陈锐 蓝贤桂[2] Lan Zhipeng;Chen Rui;Lan Xiangui(School of Mechanical and Electronic Engineering,East China University of Technology,Nanchang 330013,China;School of Information Engineering,East China University of Technology,Nanchang 330013,China)

机构地区：[1]东华理工大学机械与电子工程学院,南昌330013 [2]东华理工大学信息工程学院,南昌330013

出　　处：《机电工程技术》2023年第4期25-29,共5页Mechanical & Electrical Engineering Technology

基　　金：江西省新能源工艺及装备工程技术研究中心开放基金项目(编号:JXNE2022-06)。

摘　　要：铁路货运车辆车身携带异物容易造成重大安全隐患,出发前必须对车辆外观进行严格检查。采用深度学习方法对异物进行智能识别对提高货检工作效率具有重要意义。针对铁路货运车辆安全检测中异物识别准确率低、漏检率高等问题,以ResNet-50为基本特征提取网络,引入K-Means算法,构建了一种以交并比(Intersection over Union,IoU)为度量的锚框聚类算法,采用自建的异常目标数据集进行了实验测试,结果发现,与传统Faster RCNN相比,改进后的算法有效地增强了深度网络模型的目标特征提取能力,提高了复杂背景下铁路货运车辆异物的识别定位精度,异物的识别漏检率降低21.3%,模型具有较强的泛化能力,对异常目标精确定位研究具有一定的参考价值。Foreign objects on railway freight vehicles can easily cause significant safety hazards,and the appearance of the vehicles must be strictly inspected before its departure.Intelligent recognition of foreign objects using deep learning methods is of great significance for improving the efficiency of cargo inspection.Aiming at the problems of low accuracy and high missed rate in foreign object recognition in railway freight vehicle safety detection,a new anchor frame clustering algorithm based on Intersection over Union(IoU)was constructed using ResNet-50 as the basic feature extraction network,and K-Means algorithm was introduced.Experimental tests on our target dataset demonstrate that compared with traditional Fast RCNN,the improved algorithm effectively enhances the target feature extraction ability of the deep network model,improves the recognition and positioning accuracy of foreign objects in railway freight vehicles under complex backgrounds,and reduces the miss rate of foreign objects detection by 21.3%.With strong generalization ability,the proposed model has certain reference value for the precise location research of foreign objects.

关 键 词：铁路 车箱 异物 Faster RCNN ResNet-50 K-MEANS 

分 类 号：U298[交通运输工程—交通运输规划与管理]