基于改进YOLOv11n的X光安检图像违禁品检测算法

打开文本图片集

关键词：X光图像；深度学习；目标检测；特征融合；违禁物品；YOLOv11n中图分类号：TN911.73-34；TP391.41 文献标识码：A 文章编号：1004-373X（2026）09-0069-10

Prohibited item detection method using X-ray security inspection image based on improvedYOLOv11n

WuCuo'，ZhongHan’，Li Junjie² （1.Schoolof InformationNetwork Security，People'sPublicSecurity UniversityofChina，Beijing1ooo38，China; 2.CriminalInvestigationBrachofLiuzhouublicSecurityDepartent，aingRlwayPublicSecurityBureau，Liuzhou545oCha）

Abstract：In viewof thelow detectionaccuracyand misseddetections caused bysignificantshape variations，overlapping and occlusionsof prohibited items inX-rayimages，aswell asithesmall sizeof such items，this paperproposes an X - ray securityinspectionimageprohibiteditemdetectionalgorithmnamedHPGD-YOLOv11n，anditisbasedontheimproved YOLOvl1n.IntheBackbonepart，thealgorithmreplacesthebackbonenetworkoforiginalYOLOvl1nwithalightweight backbonenetwork HGNetV2-l，whichreducesthenumberofparametersandcomputationalloadbyoptimizingtheconvolutional structurewhilemaintaining featureextractioncapability.IntheNeck part，amoduleP-GELANisintroduced.Thismodule effectivelycapturesthedetailsofsmallobjectsbyconcatenating shallowanddepfeature maps，andenhances themulti-scale featurefusionabilitywhilereducingthenumberofparameters.IntheHeadpart，theextractioncapabilityofthemodelformultidimensionalfeaturesanditsabilitytodynamicallyadjusttodiferentinputshavebeenenhancedbyadjustingthedetectionhead of theoriginalYOLOv11ntoaDynamic Head.Experiments wereperformedonthedatasetSIXray.Theexperimentsshowthat the mean averageprecision （mAP @0.5 ）ofthe improved algorithm reaches 94.2%，and itsmAP@0.5：0.95 reaches 72.3%， whichis 1.1% and 2.10% higher than those ofthe original algorithm YOLOv11n.In addition，its parameter quantityis reduced by 19.2% .The proposed method achieves lightweight design while maintaining high detection accuracy，so it demonstrates excellent effectiveness and practical value.

Keywords：X-ray image; deep learning; object detection; feature fusion;prohibited item;YOLOv11n

0 引言

近年来，随着公共安全需求的增加，X光安检技术在机场、火车站和地铁等场所发挥着不可或缺的作用，对公共安全领域研究的进步和平安中国的建设发挥了举足轻重的作用1。（剩余15741字）