基于改进YOLOv9s的受电弓弓角检测技术

打开文本图片集

中图分类号：U279.3 文献标志码：A 文章编号：1671-5276（2025）03-0272-04

Pantograph Horn Detection Technology Based on Improved YOLOv9s

DAI Binggang1，YANG Qi1，CHEN Jianglong²， ZHANG Juhui² （1. CRRC Guangzhou Rail Transit Equipment Co.，Ltd.，Guangzhou 511495， China; （2.NanjingUniversity of Science and Technology，Nanjing 21OO94，China） Abstract：Aimedat thepoorreal-time performanceand dificult equipmentdeplymentof horn detection incomplex scenes，the traditionalConvbeingreplacedbythelightweightconvolutional GhostConvbasedonYOLOv9s，theORPEA-RepN4 moduleis designedto improvethe feature extraction module，andtheCBAM moduleisadded toenhance detectionperformance.Thetest results show that compared with YOLOv9s，the proposed modelreduces thenumberof parameters of pantograph horn detection by （204号 32.8% ，and increases the detection accuracy of horn damage by 2.3 percent，which improves horn detection performance，and meets the requirements of edge device deployment and detection tasks.

Keywords：object detection；YOLOv9s；lightweight；attention mechanism

0 引言

地铁具有节能环保、准点便捷的特点，在缓解交通拥挤和节能环保等方面有着明显的优势[1]。（剩余4644字）