改进TransCenter的组合距离多目标跟踪方法

doi:10.3778/j.issn.1002-8331.2401-0086

摘要/Abstract

摘要： 在智能驾驶和视频监控领域中，多目标跟踪被广泛应用，但在目标发生遮挡和非线性运动时，此时产生的噪声会造成检测和跟踪精度的降低，同时众多关联匹配算法也没有考虑到IoU和外观失衡的情况。针对以上问题，提出一种基于TransCenter改进的多目标跟踪网络。引用小波变换处理检测特征，设计了上下文协同选择器，通过动态选择跟踪特征和检测特征来缓解噪声产生的负面影响；融合卡尔曼滤波预测值和跟踪位移，以提高非线性运动中的预测位移准确度；根据IoU距离和外观距离的差值优化组合距离的权重，解决了高速运动和外观剧烈变化时组合距离失效的情况。在BDD100k、DanceTrack数据集上进行了实验，结果表明，与ByteTrack算法相比，改进网络的mMOTA和HOTA值分别提升了4.3和5.9个百分点，与TransCenter相比，HOTA提升了7.4个百分点，且有着更好的灵活性和跟踪精度。

关键词: 多目标跟踪, 卡尔曼滤波, 上下文协同, 组合距离

Abstract: In the fields of intelligent driving and video surveillance, multi-target tracking is widely used. However, when the target is blocked and moves non-linearly, the noise generated at this time will reduce the detection and tracking accuracy. At the same time, many correlation matching algorithms do not take this into account the IoU and appearance imbalance situations. In response to the above problems, an improved multi-target tracking network based on TransCenter is proposed. Firstly, wavelet transform is used to process detection features, and then a contextual collaborative selector is designed to alleviate the negative impact of noise by dynamically selecting tracking features and detection features. Then, Kalman filter prediction values and tracking displacements are integrated to improve nonlinear motionand the accuracy of the predicted displacement. Finally, the weight of the combined distance is optimized based on the difference between the IoU distance and the appearance distance, which solves the problem of combined distance failure when high-speed motion and dramatic changes in appearance occur. Experiments are conducted on the BDD100k and DanceTrack datasets. The results show that compared with the ByteTrack algorithm, the HOTA and mMOTA values of the improved network increase by 4.3 and 5.9 percentage points respectively. Compared with TransCenter, the HOTA increases by 7.4 percentage points and has better performance of good flexibility and tracking accuracy.

Key words: multi-target tracking, Kalman filtering, context collaboration, combined distance

赵海涛, 岳希, 唐聃, 蔡博. 改进TransCenter的组合距离多目标跟踪方法[J]. 计算机工程与应用, 2025, 61(9): 177-185.

ZHAO Haitao, YUE Xi, TANG Dan, CAI Bo. Combined Distance Multi-Target Tracking Method of Improved TransCenter[J]. Computer Engineering and Applications, 2025, 61(9): 177-185.

参考文献

[1] 张瑶, 卢焕章, 张路平, 等. 基于深度学习的视觉多目标跟踪算法综述 [J]. 计算机工程与应用, 2021, 57(13): 55-66.
ZHANG Y, LU H Z, ZHANG L P. Overview of visual multi-object tracking algorithms with deep learning[J]. Computer Engineering and Applications, 2021, 57(13): 55-66.
[2] RAKAI L, SONG H, SUN S, et al. Data association in multiple object tracking: a survey of recent techniques[J]. Expert Systems with Applications, 2022, 192: 116300.
[3] XU Y, BAN Y, DELORME G, et al. TransCenter: Transformers with dense representations for multiple-object tracking [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 45(6): 7820-7835.
[4] DENDORFER P, REZATOFIGHI H, MILAN A, et al. MOT20: a benchmark for multi object tracking in crowded scenes[J]. arXiv:2003.09003, 2020.
[5] WOJKE N, BEWLEY A, PAULUS D. Simple online and realtime tracking with a deep association metric[C]//Proceedings of the 2017 IEEE International Conference on Image Processing (ICIP), 2017.
[6] KALMAN R E. Contributions to the theory of optimal control[J]. Contributions to the Theory of Optimal Control, 1960, 5(2): 102-119.
[7] ZHANG Y, SUN P, JIANG Y, et al. ByteTrack: multi-object tracking by associating every detection box[C]//Proceedings of the 17th European Conference on Computer Vision(ECCV 2022), 2022.
[8] SUN P, CAO J, JIANG Y, et al. DanceTrack: multi-object tracking in uniform appearance and diverse motion[C]//Proceedings of the he IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022.
[9] YU F, CHEN H, WANG X, et al. BDD100k: a diverse driving dataset for heterogeneous multitask learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020.
[10] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016.
[11] WANG W, XIE E, LI X, et al. PVT v2: improved baselines with pyramid vision transformer[J]. Computational Visual Media, 2022, 8(3): 415-424.
[12] KUHN H W. The Hungarian method for the assignment problem [J]. Naval Research Logistics (NRL), 2005, 52(1): 7-21.
[13] ZHOU X, WANG D, KR?HENBüHL P. Objects as points [J]. arXiv:1904.07850, 2019.
[14] TIAN Z, SHEN C, CHEN H, et al. FCOS: a simple and strong anchor-free object detector[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 44(4): 1922-1933.
[15] SI G, ZHOU F, ZHANG Z, et al. Tracking multiple zebrafish larvae using YOLOv5 and deepsort[C]//Proceedings of the 2022 8th International Conference on Automation, Robotics and Applications (ICARA), 2022.
[16] ZENG F, DONG B, ZHANG Y, et al. MOTR: end-to-end multiple-object tracking with transformer[C]//Proceedings of the 17th European Conference on Computer Vision(ECCV 2022), 2022.
[17] SUN S, AKHTAR N, SONG H, et al. Deep affinity network for multiple object tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(1): 104-119.
[18] DU Y, ZHAO Z, SONG Y, et al. StrongSORT: make deepsort great again[J]. IEEE Transactions on Multimedia, 2023,25:8725-8737.
[19] AHARON N, ORFAIG R, BOBROVSKY B Z. BoT-SORT: robust associations multi-pedestrian tracking[J]. arXiv:2206.14651, 2022.
[20] STADLER D, BEYERER J. An improved association pipeline for multi-person tracking[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023.
[21] LOUIZOS C, WELLING M, KINGMA D P. Learning sparse neural networks through L0 regularization [J]. arXiv:1712.01312, 2017.
[22] HE L, LIAO X, LIU W, et al. Fastreid: a pytorch toolbox for general instance re-identification[C]//Proceedings of the 31st ACM International Conference on Multimedia, 2023: 9664-9667.
[23] LUO H, GU Y, LIAO X, et al. Bag of tricks and a strong baseline for deep person re-identification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2019.
[24] ZHANG H, WU C, ZHANG Z, et al. ResNeSt: split-attention networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022.
[25] 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.
[26] PAN Z, ZHUANG B, HE H, et al. Less is more: pay less attention in vision transformers[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2022.
[27] PANG J, QIU L, LI X, et al. Quasi-dense similarity learning for multiple object tracking[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021.
[28] YAN B, JIANG Y, SUN P, et al. Towards grand unification of object tracking[C]//Proceedings of the European Conference on Computer Vision, 2022.
[29] ZHOU X, KOLTUN V, KR?HENBüHL P. Tracking objects as points[C]//Proceedings of the 16th European Conference on Computer Vision(ECCV 2020), 2020.
[30] ZHANG Y, WANG C, WANG X, et al. FairMOT: on the fairness of detection and re-identification in multiple object tracking[J]. International Journal of Computer Vision, 2021, 129: 3069-3087.
[31] SUN P, CAO J, JIANG Y, et al. TransTrack: multiple object tracking with transformer [J]. arXiv:2012.15460, 2020.