基于改进ByteTrack与YOLOv10的无人机多目标跟踪算法

张忠民; 叶聪

doi:10.12382/bgxb.2025.0609

您当前的位置：

首页 >

文章列表页 >

基于改进ByteTrack与YOLOv10的无人机多目标跟踪算法

更新时间：2026-04-24

- 基于改进ByteTrack与YOLOv10的无人机多目标跟踪算法
- Drone Multi-object Tracking Algorithm Based on Improved ByteTrack and YOLOv10
- 兵工学报 2026年47卷第1期页码：250609
- 作者机构：
  
  哈尔滨工程大学信息与通信工程学院，黑龙江哈尔滨 150001
- 作者简介：
  
  *通信作者邮箱：yecong@hrbeu.edu.cn
- 基金信息：
- DOI：10.12382/bgxb.2025.0609
  中图分类号： TP391.41;V279+.2
- 收稿：2025-07-08，
  
  网络首发：2026-02-11，
  
  纸质出版：2026-01-31
- 稿件说明：
移动端阅览
张忠民, 叶聪. 基于改进ByteTrack与YOLOv10的无人机多目标跟踪算法[J]. 兵工学报, 2026,47(1):250609.

ZHANG Zhongmin, YE Cong. Drone Multi-object Tracking Algorithm Based on Improved ByteTrack and YOLOv10[J]. Acta Armamentarii, 2026, 47(1): 250609.
张忠民, 叶聪. 基于改进ByteTrack与YOLOv10的无人机多目标跟踪算法[J]. 兵工学报, 2026,47(1):250609. DOI： 10.12382/bgxb.2025.0609.

ZHANG Zhongmin, YE Cong. Drone Multi-object Tracking Algorithm Based on Improved ByteTrack and YOLOv10[J]. Acta Armamentarii, 2026, 47(1): 250609. DOI： 10.12382/bgxb.2025.0609.

摘要

无人机多目标跟踪技术是无人机领域的一个重要研究方向，目前大多数多目标跟踪技术难以平衡跟踪任务的精度和实时性。针对此问题，设计小目标检测算法MT-YOLOv10，采用轻量化特征融合模块MSKFF（Multi-Selective Kernel Feature Fusion）增强特征融合效果，提升无人机空中检测能力。在跟踪算法中，通过向卡尔曼滤波引入自适应因数增强对噪声的自适应能力，同时改变卡尔曼滤波输入的状态向量以及引入轨迹置信度信息，提升对目标位置的预测能力，改进后的跟踪算法命名为PAC-ByteTrack。MT-YOLOv10在VisDrone2019-DET数据集上的检测实验结果显示其精度和mAP50较基线算法提升4. 3%和6. 5%。将MT-YOLOv10和PAC-ByteTrack相结合，在VisDrone2019-MOT（Multi-Object Tracking）和UAVDT两大无人机数据集上展开测评，其HOTA（Harmonized Overlap and Tracking Aumulator）分别提升4. 467%和1. 831%，性能优于大多数现有跟踪算法。新算法实现了稳定连续的跟踪，为无人机跟踪任务提供了新的解决方案。

Abstract

Multi-object tracking technology for unmanned aerial vehicles (UAV) is a significant area of research in this field. At present

the majority of multi-object tracking technologies are unable to achieve an equilibrium between tracking task accuracy and real-time performance. In order to address the issue under consideration

the small object detection algorithm MT-YOLOv10 has been designed to employ a lightweight feature fusion module (MSKFF) with a view to enhancing feature fusion performance and improving the detection capabilities of UAVs in aerial contexts. In the tracking algorithm

the adaptive capability to

noise is enhanced by introducing an adaptive factor into the Kalman filter. Concurrently

modifications are made to the state vector input to the Kalman filter

with trajectory confidence information being incorporated. This process serves to enhance the prediction capability of the target posiion. The advanced tracking algorithm is designated PAC-ByteTrack. The experimental findings from MT-YOLOv10 on the VisDrone2019-DET dataset illustrate enhancements of 4. 3% and 6. 5% in accuracy and

mAP

correspondingly

in comparison with baseline algorithm. The present study combined MT-YOLOv10 with PAC-ByteTrack

conducting evaluations on the VisDrone2019-MOT and UAVDT drone datasets. The HOTA demonstrated a marked improvement

with respective increases of 4. 467% and 1. 831%

thereby surpassing the performance of the majority of existing tracking algorithms. This algorithm has been demonstrated to achieve stable and continuous tracking

thus offering a novel solution for unmanned aerial vehicle tracking tasks.

关键词

Keywords

references

WEN J Y, WANG D W, FANG J, et al. Multi-object tracking for unmanned aerial vehicles based on multi-frame feature fusion[C]∥Proceedings of International Conference on Acoustics, Speech and Signal Processing. Seoul, Korea:IEEE,2024:4180-4184.

HUANG K M, HAO Q. Joint multi-object detection and tracking with camera-LiDAR fusion for autonomous driving [C]∥Proceedings of 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems. Prague, Czech Republic: IEEE, 2021:6983-6989.

CAO W, LI Q W. A novel joint multi-target detection and tracking approach based on Bayes joint decision and estimation[J]. EURASIP Journal on Advances in Signal Processing,2023,2023（1）:71.

CHEN B, HU S L, DONG G, et al. Continuously learning video-level object tokens for robust UAV tracking[C]∥Proceedings of International Conference on Acoustics, Speech and Signal Processing. Hyderabad, India:IEEE,2025.

LIU Y Q, LI W R, DING Y H, et al. PDTrack:progressive distance association for multiple object tracking [C]∥Proceedings of International Conference on Acoustics, Speech and Signal Processing. Hyderabad, India:IEEE,2025.

DU Y H, ZHAO Z C, SONG Y, et al. Strongsort:make deepsort great again[J]. IEEE Transactions on Multimedia,2023,25:8725-8737.

WOJKE N, BEWLEY A, PAULUS D. Simple online and realtime tracking with a deep association metric[C]∥Proceedings of 2017 IEEE International Conference on Image Processing. Beijing, China:IEEE,2017:3645-3649.

CAO J K, PANG J M, WENG X S, et al. Observation-centric sort:rethinking sort for robust multi-object tracking:arXiv:2203. 14360 [R]. Ithaca, NY, US:Cornell University,2022:2203. 14360.

YANG M Z, HAN G X, YAN B, et al. Hybrid-sort: weak cues matter for online multi-object tracking: arXiv:2308. 00783 [R]. Ithaca, NY, US:Cornell University,2023:2308. 00783.

ZHANG Y F, SUN P Z, JIANG Y, et al. Bytetrack:multi-object tracking by associating every detection box[C]∥Proceedings of European Conference on Computer Vision. Cham, Switzerland:Springer,2022:1-21.

WANG A, CHEN H, LIU L H, et al. YOLOv10:real-time end-to-end object detection [J]. Advances in Neural Information Processing Systems,2024,37:107984-108011.

ZAMIR S W, ARORA A, KHAN S, et al. Learning enriched features for fast image restoration and enhancement [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2022, 45（2）:1934-1948.

LI Y B, JIU M Y, SUN Q, et al. An improved target tracking algorithm based on extended kalman filter for uav [C]∥Proceedings of 2018 IEEE Asia-Pacific Conference on Antennas and Propagation. Auckland, New Zealand:IEEE,2018:435-437.

TIAN F, WANG S Y, FU W B, et al. Vehicle target tracking algorithm based on improved strong tracking unscented Kalman filter[J]. Applied Sciences,2025,15（6）:3276.

ZHANG G W, YIN J Y, DENG P, et al. Achieving adaptive visual multi-object tracking with unscented Kalman filter[J]. Sensors, 2022,22（23）:9106.

LIU Y Y, LI Y, XU D Z, et al. Adaptive Kalman filter with power transformation for online multi-object tracking [J]. Multimedia Systems,2023,29（3）:1231-1244.

周翰祺，方东旭，张宁波，等.基于深度学习的多无人机多目标跟踪[J].计算机工程，2025,51（4）:57-65.

ZHOU H Q, FANG D X, ZHANG N B, et al.Multi-drone multi-target tracking based on deep learning [J].Computer Engineering,2025,51（4）:57-65.（in Chinese）

陈云芳，方倩，吕尊威，等.关联策略多特征增强的多目标跟踪[J/OL].计算机科学，1 - 9 [2025-06-07].https:∥link.

cnki.net/urlid/50.1075.TP.20250317.1053.009.CHEN Y F, FANG Q, LÜ Z W, et al.Multi-object tracking with multi-feature enhancement of correlation strategies [J/OL].Computer Science.1-9 [2025-06-07].https:∥link.cnki.net/urlid/50.1075.TP.20250317.1053.009.（in Chinese）

AHARON N, ORFAIG R, AND BOBROVSKY B Z. Bot-sort:robust associations multi-pedestrian tracking:arXiv:2206. 14651 [R]. Ithaca, NY, US:Cornell University,2022:2206. 14651.

LUITEN J, OSEP A, DENDORFER P, et al. Hota:a higher order metric for evaluating multi-object tracking [J]. International Journal of Computer Vision,2021,129:548-578.

WANG Z N, FENG Y, WANG L J, et al. S-YOLO:an enhanced small object detection method based on adaptive gating strategy and dynamic multi-scale focus module [J]. Neural Networks, 2025,191:107782.

CHEN X, LIN C. EVMNet: eagle visual mechanism-inspired lightweight network for small object detection in UAV aerial images[J]. Digital Signal Processing,2025,158:104957.

浏览量

下载量

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

低空复杂空间下自适应交替双目标偏差RRT^*无人机三维路径规划

基于GOTDBO算法的复杂约束条件下无人机航迹规划

基于动态特征的海上弹着点水柱检测跟踪算法

内嵌飞行动力学的高动态飞行器惯性基深耦合导航方法