首页|基于Yolo的目标识别技术

基于Yolo的目标识别技术

扫码查看
针对交通标志识别这一应用领域的性能提升,既可以提高无人驾驶系统、智能导航系统与辅助驾驶系统的性能,也可以为其他小目标检测技术提供参考.本文使用yolov8框架下的yolov8s、yolov8n与yolov8m在中国交通数据集CCTSDB基础上进行训练与性能检测,yolov8s相比于目前广泛用于目标检测的yolov5s,其mAP50提升了3.1%,且其检测速度达到了实时性要求,因此,基于yolov8s的交通标志识别可以在满足检测实时性需求的基础上提升检测的准确性.
Yolo-based Target Recognition Technology
The performance improvement of traffic sign recognition can not only improve the performance of unmanned driving systems,intelligent navigation systems and assisted driving systems,but also provide reference for other small target detection technologies.Compared with yolov5s,which is widely used for object detection,yolov8s,the mAP50 of yolov8s is improved by 3.1%,and the detection speed meets the real-time requirements,so the traffic sign recognition based on yolov8s can improve the detection accuracy on the basis of meeting the real-time requirements of detection.

traffic signsmall target recognitionunmanned drivingdriver assistanceintelligent navigation

袁巍、孟凡军

展开 >

航空工业北京航空精密机械研究所,北京 100076

交通标志 小目标识别 无人驾驶 辅助驾驶 智能导航

2024

航空精密制造技术
北京航空精密机械研究所

航空精密制造技术

CSTPCD
影响因子:0.228
ISSN:1003-5451
年,卷(期):2024.60(4)