基于改进YOLOv5模型的人车冲突概率评估
摘要
YOLOv5的改进模型。该模型首先利用 YOLOv5和DeepSORT技术识别车辆与行人之间的相对位置并进行轨迹跟
踪;其次,通过 LPET方法判别行人与车辆的冲突程度;最后,采用 SPSS进行数据分析,验证模型的有效性并计算
人车冲突概率。数据分析结果表明,模型的显著性水平 P=0.000,小于常规显著性水平,证明该冲突模型能够有效识
别无信控交叉口的人车冲突。与 YOLOv5模型对比,改进模型的准确率提升了 0.07%,表明其在无信控交叉口人车冲
突识别上具备更高的精度。
关键词
全文:
PDF参考
[1]唐鹏,陈鹏,闫伟曦.无信控路段行人与非机
动车过街行为分析.武汉理工大学学报.2022;44(2):
21-7.
TangP, ChenP, YanWX. Analysis of Pedestrian and
Non-motor Vehicle Crossing Behaviorsat Uncontrolled
Road Sections. Journal of Wuhan University of Technology.
2022; 44(2): 21-27.
[2]杨澜,赵祥模,吴国垣,等.智能网联汽车协同
生态驾驶策略综述 [J].交通运输工程学报,2020,20(5):
58-72.
[3]WojkeN, BewleyA, PaulusD. Simple online and
realtime tracking with a deep association metric[C]//2017IEEE
international conference on image processing(ICIP). IEEE,
2017: 3645-3649.
[4]余以春,李明旭.改进 YOLOv5s的自动驾驶汽车
目标检测.计算机系统应用,2023,32(9):97-105.
[5]韩晓冰,王雨田,黄综浏,等.改进 YOLOv5
+DeepSORT的行人跟踪算法 [J].现代电子技术,2023,
46(7):33-38.
[6]于浩.基于 YOLOv5+DeepSORT的多目标算法
[J].信息技术与信息化,2023(6):87-90.
[7]ZhengL, WenC, GuoY, et al. Investigating consecutive
conflicts of pedestrian crossing at unsignalized crosswalks using
the bivariate logistic approach[J].AccidentAnalysis&Prevention
,2021,162:106402.
[8]ArunA, HaqueMM, WashingtonS, et al. How many
are enough: Investigating the effectiveness of multiple conflict
in dicators for crashfrequency-by-severity estimation by
automated traffic conflict analysis[J]. Transportation research
part C: Emerging Technologies, 2022, 138: 103653.
[9]AlmodferR, hongS, FangZ, et al. Quantitative analysis
of lane-based pedestrian-vehicle conflict atanon-signalized
marked crosswalk[J]. Transportation research part F: Traffic
Psychology and Behaviour, 2016, 42: 468-478.
[10] I smailKA. Application of compute r vi sion
techniques for automated road safety analysis and traffic data
collection[D]. University of British Columbia, 2010.
Refbacks
- 当前没有refback。