基于改进YOLOv7的机场行李自动化分类方法  

Research on Automatic Baggage Classification in Airports Based on I mproved YOLOv7

作　　者：王欣[1] 陈纪宗 李屹 刘一[2] WANG Xin;CHEN Ji-zong;LI Yi;LIU Yi(School of Computer Science,Civil Aviation Flight University of China,Guanghan Sichuan 618307,China;Department of Big Data and Artificial Inte lligence,Civil Aviation Management Institut of China,Beijing 100102,China)

机构地区：[1]中国民用航空飞行学院计算机学院,四川广汉618307 [2]中国民航管理干部学院大数据与人工智能系,北京100102

出　　处：《计算机仿真》2024年第7期89-93,507,共6页Computer Simulation

基　　金：国家自然科学基金民航联合基金重点项目(U2033213,U2033214);中央高校基本科研业务费专项资金项目(J2022-048);四川省科技厅重点研发项目(2022YFG0027);中国民用航空飞行学院科研基金面上项目(J2019-045);民航飞行技术与飞行安全重点实验室飞行技术专题项目(FZ2022ZX08)。

摘　　要：针对机场旅客托运行李多样化、非规格化的特点,存在行李类别难分类,托盘回收效率低等问题,提出一种基于深度学习的目标检测算法YOLOv7对机场行李进行自动化分类的方法,通过改进后的算法来检测机场是否使用行李托盘以及对行李进行分类与定位,后续根据HSV颜色模型过滤图像背景并计算图像中白色区域面积来判定目标所属色系。最终实验结果表明,改进后的YOLOv7在本实验中针对行李类别和是否使用托盘的同时检测精度可达98.7%,色彩判定精度83%。Aiming at the characteristics of diversified and non-specified baggage checked in by airport passengers,there are problems such as difficult classification of baggage categories and low efficiency of tray recovery.A deep learning-based target detection algorithm YOLOv7 is proposed to automatically clas sify airport baggage.The improved algorithm is used to detect automatic ally whether the airprt is using baggage trays and classify and locate the baggage.And subsequently,filter the image background based on the HSV color model and ca lculate the area of the white area in the image to determine the color scheme of the target..The final experimental results show that the improved YOLOv7 can detect 98.7%of the baggage category and whether or not the tray is used at the same time,and 83%of the colour determination accuracy in this experiment.

关 键 词：智慧民航 目标检测 行李分类 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]