水下声呐图像存在背景噪声严重等问题，导致水下分类器输出大量检测低置信度对象。而现有的水下多目标跟踪框架大多简单排除低置信度目标，导致跟踪轨迹中断。本文提出一种低检测置信度下水下多目标跟踪算法YOLO-Fair MOT；引入多通道随机混合注意力模块，抑制背景噪声的影响；采用深度可分离卷积降低模型复杂性，提升跟踪过程的整体速度；结合低置信度数据匹配算法与广义交并比匹配算法，改善跟踪轨迹的中断问题。实验结果表明，YOLO-Fair MOT算法具有更好的跟踪准确度、跟踪精确度、轨迹保持性以及检测速度。
Underwater forward-looking sonar images face problems like severe background noise, leading to a large number of low confidence detection targets from the underwater classifier. Most existing underwater multi-object tracking frameworks simply exclude those targets, which cause trajectory interruption problem. A light underwater multi-target tracking algorithm YOLO-Fair MOT is proposed. The channel-spatial random fusion attention module is introduced to suppress the influence of image background noise. Depthwise convolutions are used to reduce model complexity and improves the overall speed of the tracking process. And the Byte association algorithm and the generalized intersection over union association algorithm are combined to improve the fragmentation of tracking trajectories. The experimental results show that YOLO-Fair MOT has higher value of tracking accuracy, tracking accuracy, trajectory retention, and detection speed.
2025,47(6): 128-133 收稿日期：2024-4-11
DOI：10.3404/j.issn.1672-7649.2025.06.021
分类号：TP391.4
基金项目：国家自然科学基金资助项目（62373285）；上海市产业协同创新项目（HCXBCY-2022-051）；机器人技术与系统全国重点实验室开放基金（SKLRS-2024-KF-04）；某部基础科研计划项目（XXXX2022YYYC133）
作者简介：张文凯（2001 – ），男，博士研究生，研究方向为水下机器人
参考文献：
[1] 郭戈, 王兴凯, 徐慧朴. 基于声呐图像的水下目标检测、识别与跟踪研究综述[J]. 控制与决策, 2018, 33(5): 906-922.
GUO G, WANG X K, XU H P. Review on underwater target detection, recognition and tracking based on sonar image[J]. Control and Decision, 2018, 33(5): 906-922.
[2] KIM B, YU S. Imaging sonar based real-time underwater object detection utilizing AdaBoost method[C]// 2017 IEEE Underwater Technology(UT), Busan, Korea, 2017.
[3] 黄松威, 朱兆彤, 胡友峰. 基于粒子滤波的声呐图像目标跟踪算法研究[J]. 舰船科学技术, 2019, 41(3): 135-139.
HUANG S W, ZHU Z T, HU Y F. A sonar image target tracking algorithm based on particle filter[J]. Ship science and technology, 2019, 41(3): 135-139.
[4] ZHAI X Y, WEI H L, WU H D, et al. Multi-target tracking algorithm in aquaculture monitoring based on deep learning[J]. Ocean Engineering, 2023, 289(1): 116005.
[5] ZHAO X H, HUANG Z and LV Y J. Research on real-time diver detection and tracking method based on YOLOv5 and DeepSORT[C]// 2022 IEEE International Conference on Mechatronics and Automation(ICMA), China, 2022.
[6] ZHANG Y F, WANG C Y, WANG X G, et al. FairMOT: on the fairness of detection and re-identification in multiple object tracking[J]. International Journal of Computer Vision, 2021, 129(11): 3069-3087.
[7] ANDREW H, ZHU M, CHEN B, et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications[J/OL]. 2017, 17(4): 48-61.
[8] ZHANG Y F, SUN P Z, JIANG Y. ByteTrack: Multi-object tracking by associating every detection box[C]// Computer Vision–ECCV 2022. Cham, Springer, 2022.
[9] Blueprint subsea. Oculus[DB/OL]. 2018, (2018-08-01)[2023-08-25]. https://www.blueprintsubsea.com/oculus/.
