央企文库

当前位置:首页 > 汇报材料 > 调研报告

调研报告

基于改进SSD 的交通标志实时检测

2020-01-01 00:00:00调研报告
徐元泽 王勇摘要:针对实际驾驶时获取的自然交通场景图像中,交通标志占图像比例较小导致检测精度低问题,同时要求交通标志检测速度快,提出一种改进的单发多框检测(SSD)神经网络模型。该模型在SSD基础上融合特征金字塔网络(FPN

徐元泽 王勇

摘要:针对实际驾驶时获取的自然交通场景图像中,交通标志占图像比例较小导致检测精度低问题,同时要求交通标志检测速度快,提出一种改进的单发多框检测(SSD)神经网络模型。该模型在SSD基础上融合特征金字塔网络(FPN),在后处理方法应用中心点距离非极大值抑制(DIoU-NMS),提高了交通标志小目标的检测精度。实验结果表明,改进后的SSD 网络型检测性能显著提高,其均值平均精度(mAP)比原SSD提高了7.6个百分点,其每秒帧率(FPS)达到31.4具备实时检测能力。

关键词:交通标志;实时检测;单发多框检测(SSD);特征金字塔网络(FPN)

中图分类号:TP183  文献标识码:A

文章编号:1009-3044(2021)29-0092-03

Real-time Traffic Signs Detection Based on Improved SSD

XU Yuan-ze,WANG Yong

(School of Computers, Guangdong University of Technology, Guangdong 51000 China )

Abstract:In the natural traffic scene images obtained during actual driving, the small proportion of traffic signs in the image leads to the problem of low detection accuracy. At the sam ……此处隐藏5000个字…… ECCV 2016.Cham:Springer In⁃ternational Publishing,2016:21-37.

[8] Redmon J,DivvalaS,GirshickR,etal.You only look once:uni⁃fied, real-time ect detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).June 27-30, 2016,Las Vegas,NV,USA.IEEE,2016:779-788.

[9] Lin T Y,DollárP,GirshickR,etal.Feature pyramid networks for ect detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).July 21-26,2017,Ho⁃ nolulu,HI,USA.IEEE,2017:936-944.

[10] Zheng Z H,WangP,LiuW,etal.Distance-IoUloss:faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7):12993-13000.

【通聯编辑:唐一东】

 

文章评论