Highway billboard inspection object tracking based on improved ByteTrack algorithm

doi:10.11947/j.AGCS.2025.20250149

Abstract

Abstract:

The ByteTrack algorithm is used to track the billboards captured by the camera of the highway inspection vehicle, which can extract the video screen of the billboard and the information of the time node. However, the algorithm faces the challenge of occlusion and false tracking of non-advertisement objects. Therefore, the following improvements are made to the ByteTrack algorithm. Firstly, a buffer trajectory needs to be created before the object is identified as the tracking ID until the trajectory satisfies the pre-activation criterion. The occlusion state is judged for the object trajectory in the lost state. When the pre-activated object and the occluded object meet the category, appearance and orientation vector conditions, the Hungarian matching between the two objects is performed. Secondly, referring to the characteristics of Kalman filter parameter setting in Botsort and ByteTrack algorithms, genetic algorithm is used to adjust the key parameters of Kalman filter under XYAH and XYWH coding modes respectively, and the Kalman filter with the best prediction effect is selected. Some expressways in Changsha-Zhuzhou-Xiangtan urban agglomeration are selected as experimental objects. The results show that compared with the original ByteTrack algorithm, the proposed method improves the Hota, Mota and IDF indexes by 1.318%, 11.682% and 2.033%, respectively. Compared with other multi-object tracking algorithms, the improved ByteTrack algorithm is superior to Botsort, Deepocsort, Hybridsort and other algorithms except that the FP value is slightly higher than the Ocsort algorithm. The improved ByteTrack algorithm achieves good tracking of object on highway billboards, which provides a reference for the intelligent inspection technology of highway billboards.

Key words: highway, inspection vehicle, billboard, ByteTrack, multi-object tracking

CLC Number:

P208

Jun LI, Chaokui LI, Lei HUANG, Yuanyuan FENG. Highway billboard inspection object tracking based on improved ByteTrack algorithm[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(11): 2068-2080.

Figures/Tables 15

Fig. 1

Tab. 1

Fig. 2

Fig. 3

Tab. 2

Tab. 3

Fig. 4

Fig. 5

Fig. 6

Tab. 4

Tab. 5

Tab. 6

Tab. 7

Fig. 7

Fig. 8

References 25

[1]	杨晶, 王丽园, 刘兴东, 等. 高速公路全生命周期智慧巡养管理系统设计与实现[J]. 公路, 2021, 66(11): 328-334.
	YANG Jing, WANG Liyuan, LIU Xingdong, et al. Design and implementation of intelligent inspection and management system in the whole life of expressway[J]. Highway, 2021, 66(11): 328-334.
[2]	党倩. 基于无人机的高速公路违法广告设施智能巡查技术应用[J]. 电子技术与软件工程, 2020(19): 123-125.
	DANG Qian. Application of intelligent inspection technology for illegal advertising facilities on expressway based on UAV[J]. Electronic Technology & Software Engineering, 2020(19): 123-125.
[3]	陈鹏武, 徐东华, 魏安, 等. 关于无人机巡检技术在高速公路中应用与研究[J]. 交通科技与管理, 2024(18): 4-6.
	CHEN Pengwu, XU Donghua, WEI An, et al. Application and research of UAV patrol technology in expressway[J]. Traffic Technology and Management, 2024, 5(18): 4-6.
[4]	朱建伟. 基于车载遥感的高速公路路域广告牌智能巡检关键技术研究[D]. 湘潭: 湖南科技大学, 2022.
	ZHU Jianwei. Research on key technologies of intelligent inspection of billboards in expressway based on vehicle remote sensing[D]. Xiangtan: Hunan University of Science and Technology, 2022.
[5]	张永泽, 达飞鹏. 膨胀区域匹配和自适应轨迹管理策略的多目标跟踪方法[J]. 武汉大学学报(信息科学版), 2024, 49(4): 572-581, 594.
	ZHANG Yongze, DA Feipeng. A multi-object tracking method based on dilatation region matching and adaptive trajectory management strategy[J]. Geomatics and Information Science of Wuhan University, 2024, 49(4): 572-581, 594.
[6]	李杰, 张洛维, 王晓燕, 等. 基于视锥距离和自适应权重卡尔曼滤波的多传感器融合算法[J]. 中国公路学报, 2024, 37(3): 194-203.
	LI Jie, ZHANG Luowei, WANG Xiaoyan, et al. A multi-sensor fusion algorithm based on view-cone distance and adaptive weighted Kalman filter[J]. China Journal of Highway and Transport, 2024, 37(3): 194-203.
[7]	李永上, 马荣贵, 张美月. 改进YOLOv5s+DeepSORT的监控视频车流量统计[J]. 计算机工程与应用, 2022, 58(5): 271-279.
	LI Yongshang, MA Ronggui, ZHANG Meiyue. Traffic monitoring video vehicle volume statistics method based on improved YOLOv5s+DeepSORT[J]. Computer Engineering and Applications, 2022, 58(5): 271-279.
[8]	李德仁, 洪勇, 王密, 等. 测绘遥感能为智能驾驶做什么?[J]. 测绘学报, 2021, 50(11): 1421-1431. DOI: 10.11947/j.AGCS.2021.20210280.
	LI Deren, HONG Yong, WANG Mi, et al. What can surveying and remote sensing do for intelligent driving?[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(11): 1421-1431. DOI: 10.11947/j.AGCS.2021.20210280.
[9]	李景文, 韦晶闪, 姜建武, 等. 多视角监控视频中动态目标的时空信息提取方法[J]. 测绘学报, 2022, 51(3): 388-400. DOI: 10.11947/j.AGCS.2022.20200507.
	LI Jingwen, WEI Jingshan, JIANG Jianwu, et al. Spatio-temporal information extraction method for dynamic targets in multi-perspective surveillance video[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(3): 388-400. DOI: 10.11947/j.AGCS.2022.20200507.
[10]	ZHANG Yifu, SUN Peize, JIANG Yi, et al. ByteTrack: multi-object tracking by associating every detection box[C]//Proceedings of 2022 ECCV. Cham: Springer, 2022: 1-21.
[11]	VEERAMANI B, RAYMOND J W, CHANDA P. DeepSort: deep convolutional networks for sorting haploid maize seeds[J]. BMC Bioinformatics, 2018, 19(9): 289.
[12]	聂源, 赖惠成, 高古学. 改进YOLOv7+Bytetrack的小目标检测与追踪[J]. 计算机工程与应用, 2024, 60(12): 189-202.
	NIE Yuan, LAI Huicheng, GAO Guxue. Improved small target detection and tracking with YOLOv7+Bytetrack[J]. Computer Engineering and Applications, 2024, 60(12): 189-202.
[13]	邹北骥, 李伯洲, 刘姝. 基于中心点检测和重识别的多行人跟踪算法[J]. 武汉大学学报(信息科学版), 2021, 46(9): 1345-1353.
	ZOU Beiji, LI Bozhou, LIU Shu. A multi-pedestrian tracking algorithm based on center point detection and person re-identification[J]. Geomatics and Information Science of Wuhan University, 2021, 46(9): 1345-1353.
[14]	KALMAN R E. A new approach to linear filtering and prediction problems[J]. Journal of Basic Engineering, 1960, 82(1): 35-45.
[15]	许华杰, 郑力文. 基于自适应卡尔曼滤波的视觉多目标跟踪[J]. 计算机工程与应用, 2025, 61(5): 200-210.
	XU Huajie, ZHENG Liwen. Visual multi-object tracking based on adaptive Kalman filter[J]. Computer Engineering and Applications, 2025, 61(5): 200-210.
[16]	AHARON N, ORFAIG R, BOBROVSKY B Z. BoT-SORT: robust associations multi-pedestrian tracking[EB/OL]. [2025-08-19]. https://arxiv.org/abs/2206.14651.
[17]	HOLLAND J H. Genetic algorithms[J]. Scientific American, 1992, 267(1): 66-72.
[18]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 770-778.
[19]	MUNKRES J. Algorithms for the assignment and transportation problems[J]. Journal of the Society for Industrial and Applied Mathematics, 1957, 5(1): 32-38.
[20]	VARGHESE R, SAMBATH M. YOLOv8: a novel object detection algorithm with enhanced performance and robustness[C]//Proceedings of 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems. Chennai: IEEE, 2024: 1-6.
[21]	BERNARDIN K, STIEFELHAGEN R. Evaluating multiple object tracking performance: the CLEAR MOT metrics[J]. EURASIP Journal on Image and Video Processing, 2008, 2008(1): 246309.
[22]	LUITEN J, OŠEP A, DENDORFER P, et al. HOTA: a higher order metric for evaluating multi-object tracking[J]. International Journal of Computer Vision, 2021, 129(2): 548-578.
[23]	MAGGIOLINO G, AHMAD A, CAO Jinkun, et al. Deep OC-sort: multi-pedestrian tracking by adaptive re-identification[C]//Proceedings of 2023 IEEE International Conference on Image Processing. Kuala Lumpur: IEEE, 2023: 3025-3029.
[24]	CAO Jinkun, PANG Jiangmiao, WENG Xinshuo, et al. Observation-centric SORT: rethinking SORT for robust multi-object tracking[C]//Proceedings of 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver: IEEE, 2023: 9686-9696.
[25]	SAMANTA R, FUNKHOUSER T, LI Kai, et al. Hybrid sort-first and sort-last parallel rendering with a cluster of PCs[C]//Proceedings of 2000 ACM Siggraph/Eurographics Workshop on Graphics Hardware. Interlaken Switzerland: ACM Press, 2000: 97-108.

区域	判定方法
左边缘	x_tl<E_W
右边缘	x_br>W-E_W
上边缘	y_tl>H-E_H

边界线	交点
x=0	（0，b）
x=W	（W，m×W+b）
y=0
y=H

编码方式	关键参数	表示方式
XYWH	x₀	[x，y，w，h，0，0，0，0]
	P₀	diag[（w/k₁）²，（h/k₁）²，（w/k₁）²，（h/k₁）²，（w/k₂）²，（h/k₂）²，（w/k₂）²，（h/k₂）²]
	Q	diag[（w/k₃）²，（h/k₃）²，（w/k₃）²，（h/k₃）²，（w/k₄）²，（h/k₄）²，（w/k₄）²，（h/k₄）²]
	R	diag[（w/k₅）²，（h/k₅）²，（w/k₅）²，（h/k₅）²]
XYAH	x₀	[x，y，a，h，0，0，0，0]
	P₀	diag[（h/k₁）²，（h/k₁）²，（0.01）²，（h/k₁）²，（h/k₂）²，（h/k₂）²，（0.000 01）²，（h/k₂）²]
	Q	diag[（h/k₃）²，（h/k₃）²，（0.01）²，（h/k₃）²，（h/k₄）²，（h/k₄）²，（0.000 01）²，（h/k₄）²]
	R	diag[（h/k₅）²，（h/k₅）²，（h/k₅）²，（h/k₅）²]

累计跟踪帧数	Hota/（%）↑	Mota/（%）↑	IDF/（%）↑	IDS↓
10	54.058	54.516	77.981	13
20	54.008	56.291	78.183	11
30	53.92	57.208	78.218	10
40	53.396	57.03	77.778	8
50	53.099	56.829	77.316	5
60	52.566	56.173	76.534	4

丢失帧数	Hota/（%）↑	Mota/（%）↑	IDF/（%）↑	IDS↓
30	53.92	57.208	78.218	10
60	53.959	57.143	78.364	5
90	53.963	57.231	78.514	3
120	53.923	57.211	78.506	3
150	53.923	57.211	78.506	3