基于深度学习的铁路集装箱装车偏载智能化监测技术  

Intelligent Monitoring Technology for Railroad Container Loading Deviation Based on Computer Vision and Deep Learning

作　　者：李建国 张双霞 杜海平 Li Jianguo;Zhang Shuangxia;Du Haiping(Railway Freight Department,China Railway Lanzhou Bureau Group Co.,Ltd.,Gansu,Lanzhou 730070,China;Key Laboratory of Railway Industry on Plateau Railway Transportation Intelligent Management and Control,Gansu,Lanzhou 730070,China)

机构地区：[1]中国铁路兰州局集团有限公司货运处,甘肃兰州730070 [2]高原铁路运输智慧管控铁路行业重点实验室,甘肃兰州730070

出　　处：《铁道技术标准(中英文)》2025年第3期1-8,共8页Railway Technical Standard(Chinese & English)

摘　　要：本文聚焦于解决敞车装载集装箱时存在的流程繁琐、效率低下以及劳动安全风险大等问题,推动敞车装载集装箱向智能化发展。通过运用YOLOv8的目标分割和关键点检测技术,制定相应判断规则,实现对集装箱偏载情况的智能监测。同时搭建可视化平台,利用其界面将装载后的信息实时反馈给门式起重机操作员,省去人工测量环节,提高装载效率,让装载过程更智能化。操作员可据此直观获取集装箱装载情况,快速准确地进行调整。基于YOLOv8算法,模型在分割和关键点检测上的mAP均达99.5%,准确完成箱体的关键点检测和箱体分割,有效解决了敞车装载集装箱偏载问题,具有重要的研究价值与实用意义。This paper focuses on solving the problems of cumbersome process,low efficiency and high risk of labor safety when loading containers in open trucks,and promoting the intelligent development of loading containers in open trucks.By applying the target segmentation and key point detection technology of yolov8 and formulating corresponding judgment rules,the project realizes the intelligent monitoring of container offloading.At the same time,a visualization platform is established,using its interface to feedback in real-time the loading information to the gantry crane operator,thus eliminating the manual measurement link and improving loading efficiency,and making the loading process more intelligent.The operator can obtain the container loading situation visually and make adjustments quickly and accurately.Based on YOLOv8 algorithm,the mAP of the model in segmentation and key point detection reaches 99.5%,which accurately completes the key point detection and segmentation of the box,and effectively solves the problem of deviation of container loading in open trucks,which is of important research value and practical significance.

关 键 词：货场集装箱 计算机视觉 深度学习 深度监督 可视化 

分 类 号：U294.2[交通运输工程—交通运输规划与管理]