基于改进YOLOv11n的复杂场景下行人检测模型

打开文本图片集

中图分类号：TP391.4 文献标识码：A DOI：10.7535/hbkd.2026yx01007

Abstract：Toaddress thedecline in pedestriandetectionaccuracycaused bycomplexscenariossuchasilumination variations， viewingangles，ackgroundinterferenceandsmallpedestriantargetswhichoftenleadtofalsepositivesandmiseddetections， a pedestrian detection model， YOLOv11-CREP，was proposed based onan improved YOLOv11n.Firstly，CSPDConv，which wasformed byintegratingstandardconvolution（Conv）withspace-to-depthconvolution（SPDConv），was introducedtoreduce information lossand enhance critical feature extraction.Secondly，a new RepNCSPELAN4-GC module was proposed，which incorporates GhostConv to optimize the RepNCSPELAN4 module，reducing its parameter count.The improved RepNCSPELAN4-GC module was then used to partiall replace theC3k2 modules in the Neck layer.Next，eficient multi-scale atention（EMAttention）andparallel network attention（ParNetAttention）werefused intoanew EMPAtentionmoduleto enhancethedetectionabilityofthemodelforsmalltargetpedestrians.Finally，consideringthecharacteristicsofsmall target pedestrainsandoccluded targets，asmal-targetdetectionheadP2wasadded tofurtherimprovethemodel’srecognition capability for small targets. The experiments show that compared with the original YOLOvl1n model，YOLOv11-CREP improves the mean average precision（mAP）by 4.6 percentage points at an IoU threshold of O.5，reaching 95.3% . When evaluated over the IoU range of O.5 to O.95，its mAP increases by 9.O percentage points，reaching 70.2% .The proposed modelachievesabalancebetweenhighdetectionperformanceandreal-timerequirements，effectivelyenhancingpedestrian detectio performance in complex scenarios.It provides valuable references for modeling pedestrian detection tasks.

Keywords：computer image processing；YOLOvlln；pedestrian detection;complex scenarios；atention mechanisms；smal object detection

行人检测是目标检测领域的重要任务，广泛应用于公共安全、自动驾驶、智能监控等场景，对保障社会安全与提升系统智能化水平具有重要意义。（剩余15874字）