基于自动驾驶的路口交通信号灯识别与应用

打开文本图片集

中图法分类号TP27 文献标识码A 文章编号：2096-4706（2025）21-0072-06

Recognition and Application of Intersection Traffic Signal Lights Based on Autonomous Driving

LUOXiaoxuan，ENGWei，XUDeheng，WANGBingkun，YUPan （Jiangxi Institute ofTechnology，Nanchang 33oo98，China）

Abstract： Trafic signal lightrecognition is oneofthecore technologies inautonomous driving.However，the performance of existing methods incomplex environmentssuchaslightchanges，smalltargets atadistance，andoclusioncan hardlymeet therequirementsofeal-tieandobuste.Forisason，thetudypropossaecienttafcsigallightTargetDtection method basedontheYOLOalgorithm.Firstly，thedatasetis preprocessd，andthen theYOLOv8and YOLOv5modelsare trained separatelyand theperformanceiscomparedbasedontheF1curve，PRcurve，andavarietyofevaluationindexes. The experimental results show that YOLOv8outperforms YOLOv5 in terms of precision，recal，mAPand other indexes， demonstratingbeterdetectionaccuracyand robustness.This study verifiesthe superiorityofYOLOv8 inthe autonomous driving signal light detection task，whichcan provide technical supportreference forsafedriving incomplex traffcscenarios.

Keywords：YOLO;signal light detection; Target Detection; data enhancement

0 引言

随着科技的发展和社会的进步，自动驾驶已成为人类探索未来出行方式的重要方式。（剩余5711字）