煤矿井下环境复杂，人员服装统一且容易被频繁遮挡，通用跨摄像头跟踪方法不能很好适用于煤矿井下场景。因此应用于煤矿井下的跨摄像头单目标跟踪方法的研究尤为重要。本文以煤矿井下单目标跟踪和行人重识别为研究对象，提出基于概率回归的煤矿井下单目标跟踪方法和基于注意力引导的煤矿井下行人重识别方法。主要工作和创新点如下：

（1）针对煤矿场景下当前单目标跟踪方法对相似目标判别性差导致跟踪漂移的问题，提出了一种基于概率回归的煤矿井下单目标跟踪方法。首先，采用高斯采样的概率回归算法，在后续帧中预测目标的位置。其次，通过学习输入与目标状态之间的条件概率密度，为具有相似标注的目标建立概率输出模型。旨在解决因目标姿态动态变化或意外遮挡所引发的不准确性问题。同时，采用ResNeSt作为特征提取的主干网络，通过设计的边界框回归机制，有效区分外观相似的目标。最后在训练过程中最小化Kullback-Leibler（KL）散度来验证架构的有效性。实验结果表明，提出方法在三个基准数据集上取得了更好的效果，AUC增加到83.7%，跟踪速度超过60FPS。

（2）针对煤矿场景下当前行人重识别方法识别精度差导致匹配错误的问题，提出了一种基于注意力引导的煤矿井下行人重识别网络。首先，采用ResNet网络来提取目标人员细节信息。其次，利用注意力诱导的跨层次融合模块创建了一个新的特征融合分支，该分支执行跨层次学习并增强了类间可比对象的表示，同时增强相似目标之间的细节特征。最后，通过具有对比性全局特征来生成强大的特征表示，降低人员间相似性的识别难度。在自建数据集MineReIdData和公共数据集Market-1501进行了实验，结果表明，提出方法取得了最优的效果，mAP分别达到了65.6%和88.3%。

（3）在以上工作基础上，设计实现了一个煤矿井下跨摄像头单目标跟踪系统，包括实时跨摄像头在线监控跟踪、摄像头设备管理、检测定时，以及用户和消息管理功能。功能和性能测试结果表明，该系统具有很好的稳定性和可靠性。

The environment in coal mines is complex, and personnel clothing is uniform and easily frequently occluded, making general cross-camera tracking methods not well applicable to underground coal mine scenarios. Therefore, the research on cross-camera single-target tracking methods applied in underground coal mines is particularly important. This article takes underground coal mine single-target tracking and pedestrian re-identification as research objects, proposing a probability regression-based underground coal mine single-target tracking method and an attention-guided underground coal mine pedestrian re-identification method. The main work and innovations are as follows:

(1) To address the issue of tracking drift caused by the poor discrimination of similar targets in current single-target tracking methods in coal mine scenarios, a probability regression-based single-target tracking method for underground coal mines is proposed. First, a probability regression algorithm using Gaussian sampling is employed to predict the target's position in subsequent frames. Second, by learning the conditional probability density between the input and the target state, a probability output model is established for targets with similar annotations. This aims to address the inaccuracy issues caused by dynamic changes in target poses or unexpected occlusions. Meanwhile, ResNeSt is adopted as the backbone network for feature extraction, and an effective distinction between visually similar targets is achieved through the designed bounding box regression mechanism. Finally, the Kullback-Leibler (KL) divergence is minimized during training to verify the effectiveness of the proposed architecture. Experimental results show that the proposed method achieves better performance on three benchmark datasets, with an AUC increase to 83.7% and a tracking speed exceeding 60 FPS.

(2) To address the issue of matching errors caused by the poor recognition accuracy of current pedestrian re-identification methods in coal mine scenarios, an attention-guided network for pedestrian re-identification in underground coal mines is proposed. First, a ResNet network is employed to extract detailed information of the target personnel. Second, a novel feature fusion branch is created using an attention-induced cross-level fusion module, which performs cross-level learning and enhances the representation of comparable objects between classes while strengthening the detailed features between similar targets. Finally, a robust feature representation is generated through contrastive global features, reducing the difficulty of identifying similarities between personnel. Experiments conducted on the self-built MineReIdData dataset and the public Market-1501 dataset show that the proposed method achieves optimal results, with mAP reaching 65.6% and 88.3%, respectively.

(3) Based on the above work, a cross-camera single-target tracking system for underground coal mines has been designed and implemented, including real-time cross-camera online monitoring and tracking, camera device management, detection timing, and user and message management functions. Functional and performance test results show that the system has excellent stability and reliability.

[1] 王红涛,邓淼磊,赵文君,等.基于深度学习的单目标跟踪算法综述[J].计算机系统应用,2022,31(05):40-51.DOI:10.15888/j.cnki.csa.008476.

[2] Ma C, Huang J B, Yang X, et al. Hierarchical convolutional features for visual tracking[C]//Proceedings of the IEEE international conference on computer vision. 2015: 3074-3082.

[3] Danelljan M, Robinson A, Shahbaz Khan F, et al. Beyond correlation filters: Learning continuous convolution operators for visual tracking[C]//Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part V 14. Springer International Publishing, 2016: 472-488.

[4] Danelljan M, Bhat G, Shahbaz Khan F, et al. Eco: Efficient convolution operators for tracking[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 6638-6646.

[5] BhatG J J. Unveilingthepower ofdeeptracking//ProceedingsoftheEuropeanConferenceon ComputerVision[J]. Munich, Germany, 2018, 493: 509.

[6] Lipton Z C, Berkowitz J, Elkan C. A critical review of recurrent neural networks for sequence learning[J]. arXiv preprint arXiv:1506.00019, 2015.

[7] Danelljan M, Bhat G, Khan F S, et al. Atom: Accurate tracking by overlap maximization[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4660-4669.

[8] Chang C, Ansari R. Kernel particle filter for visual tracking[J]. IEEE signal processing letters, 2005, 12(3): 242-245.

[9] Bertinetto L, Valmadre J, Henriques J F, et al. Fully-convolutional siamese networks for object tracking[C]//Computer Vision–ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part II 14. Springer International Publishing, 2016: 850-865.

[10] Li B, Wu W, Wang Q, et al. Siamrpn++: Evolution of siamese visual tracking with very deep networks[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4282-4291.

[11] Li B, Wu W, Wang Q, et al. Siamrpn: Tracking Objects as Points[C]// In Proceedings of the European Conference on Computer Vision (ECCV). 2019: 484-500

[12] 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: 770-778.

[13] Wang Q, Zhang L, Bertinetto L, et al. Fast online object tracking and segmentation: A unifying approach[C]//Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. 2019: 1328-1338.

[14] Yu Y, Xiong Y, Huang W, et al. Deformable siamese attention networks for visual object tracking[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 6728-6737.

[15] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30.

[16] Chen X, Yan B, Zhu J, et al. Transformer tracking[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 8126-8135.

[17] He K, Zhang C, Xie S, et al. Target-aware tracking with long-term context attention[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2023, 37(1): 773-780.

[18] Bhat G, Danelljan M, Gool L V, et al. Learning discriminative model prediction for tracking[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2019: 6182-6191.

[19] 吴德政.数字化矿山现状及发展展望[J].煤炭科学技术.2014,42(09):17-21.

[20] 原磊明,王海斌,刘旭飞,等.矿井人身安全视频防护系统设计[J].煤炭技术,2018,37(11):373-375.

[21] 贾澎涛，贾伟.煤矿井下视频多目标轨迹跟踪算法研究[J].计算机工程与应用,2018,54(02):222-227.

[22] 郝建华.基于Cam Shift和粒子滤波的煤矿变电所人员目标踪算法研究[J].工矿自动化,2015,41(11):35-38.

[23] 崔丽珍,邬嵩,赵晓燕,等.面向煤矿井下的改进分布式目标跟踪算法[J].煤矿安全,2017,48(02):113-116.

[24] 孙继平,贾倪.矿井视频图像中人员目标匹配与跟踪方法[J].中国矿业大学学报,2015,44(03):540-548.

[25] 陈伟,任鹏,田子建,等.基于注意力机制的无监督矿井人员跟踪[J/OL].煤炭学报:1-8[2021-05-24].https://doi.org/10.13225/j.cnki.jccs.2020.1348.

[26] 程德强,刘洁,郭政.基于分层光流的煤矿井下运动目标跟踪算法[J].工矿自动化,2015,41(03):75-79.

[27] 耿蒲龙,宋建成,刘旭飞,等.基于RGB颜色空间的矿井运动目标检测及跟踪方法[J].太原理工大学学报,2017,48(06):963-968.

[28] 陈亮.煤矿井下复杂场景动态目标跟踪算法研究[D].中国矿业大学,2021.DOI:10.27623/d.cnki.gzkyu.2021.000754.

[29] 黄业文,许智聪,单纯.基于深度学习的行人重识别综述[J].广州大学学报(自然科学版),2022,21(02):30-41.

[30] Sun Y, Zheng L, Deng W, et al. Svdnet for pedestrian retrieval[C]//Proceedings of the IEEE international conference on computer vision. 2017: 3800-3808.

[31] Hermans A, Beyer L, Leibe B. In defense of the triplet loss for person re-identification[J]. arXiv preprint arXiv:1703.07737, 2017.

[32] Li W, Zhu X, Gong S. Harmonious attention network for person re-identification[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 2285-2294.

[33] Bai X, Yang M, Huang T, et al. Deep-person: Learning discriminative deep features for person re-identification[J]. Pattern Recognition, 2020, 98: 107036.

[34] Zhang X, Luo H, Fan X, et al. Alignedreid: Surpassing human-level performance in person re-identification[J]. arXiv preprint arXiv:1711.08184, 2017.

[35] Sun Y, Zheng L, Yang Y, et al. Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline)[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 480-496.

[36] Wang G, Yuan Y, Chen X, et al. Learning discriminative features with multiple granularities for person re-identification[C]//Proceedings of the 26th ACM international conference on Multimedia. 2018: 274-282.

[37] Fu Y, Wei Y, Zhou Y, et al. Horizontal pyramid matching for person re-identification[C]//Proceedings of the AAAI conference on artificial intelligence. 2019, 33(01): 8295-8302.

[38] Park H, Ham B. Relation network for person re-identification[C]//Proceedings of the AAAI conference on artificial intelligence. 2020, 34(07): 11839-11847.

[39] Peng Z, Huang W, Gu S, et al. Conformer: Local features coupling global representations for visual recognition[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 367-376.

[40] 陈璠,彭力.异构分支关联特征融合的行人重识别[J].计算机科学与探索,2022,16(11):2609-2618.

[41] 钱亚萍,王凤随,熊磊.基于局部细化多分支与全局特征共享的无监督行人重识别方法[J].电子测量与仪器学报,2023,37(01):106-115.DOI:10.13382/j.jemi.B2205837.

[42] 郑烨,赵杰煜,王翀,等.基于姿态引导对齐网络的局部行人再识别[J].计算机工程,2020,46(05):247-253.DOI:10.19678/j.issn.1000-3428.0056642.

[43] Ye M, Lan X, Wang Z, et al. Bi-directional center-constrained top-ranking for visible thermal person re-identification[J]. IEEE transactions on information forensics and security, 2019, 15: 407-419.

[44] Qin D, Gammeter S, Bossard L, et al. Hello neighbor: Accurate object retrieval with k-reciprocal nearest neighbors[C]//CVPR 2011. IEEE, 2011: 777-784.

[45] 周智恒,刘楷怡,黄俊楚,等.一种基于等距度量学习策略的行人重识别改进算法[J].电子与信息学报,2019,41(02):477-483.

[46] 章东平,徐丽园.距离度量学习的摄像网络中行人重识别[J].中国计量大学学报,2016,27(04):424-428+434.

[47] 张立亚,王寓,郝博南.基于改进度量学习的煤矿井下行人重识别方法研究[J].工矿自动化,2023,49(09):84-89+166.DOI:10.13272/j.issn.1671-251x.18100.

[48] Fan H, Lin L, Yang F, et al. Lasot: A high-quality benchmark for large-scale single object tracking[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 5374-5383.

[49] Wu Y, Lim J, Yang M H. Online object tracking: A benchmark[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2013: 2411-2418.

[50] Zheng L, Shen L, Tian L, et al. Scalable person re-identification: A benchmark[C]//Proceedings of the IEEE international conference on computer vision. 2015: 1116-1124.

[51] 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: 2736-2746.

[52] 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: 770-778.

[53] Danelljan M, Gool L V, Timofte R. Probabilistic regression for visual tracking[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 7183-7192.

[54] Danelljan M, Bhat G, Khan F S, et al. Atom: Accurate tracking by overlap maximization[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4660-4669.

[55] Huang L, Zhao X, Huang K. Got-10k: A large high-diversity benchmark for generic object tracking in the wild[J]. IEEE transactions on pattern analysis and machine intelligence, 2019, 43(5): 1562-1577.

[56] Kingma D P, Ba J. Adam: A method for stochastic optimization[J]. arXiv preprint arXiv:1412.6980, 2014.

[57] Yao F, Wang S, Ding L, et al. Attention-Guided Multi-Scale Fusion Network for Similar Objects Semantic Segmentation[J]. Cognitive Computation, 2024, 16(1): 366-376.

[58] Wang S, Cao J, Philip S Y. Deep learning for spatio-temporal data mining: A survey[J]. IEEE transactions on knowledge and data engineering, 2020, 34(8): 3681-3700.

[59] Ketkar N, Ketkar N. Stochastic gradient descent[J]. Deep learning with Python: A hands-on introduction, 2017: 113-132.