[1]林志鸿,郑力新,曾远跃.采用空间依赖的MTDPN扶梯危险行为的姿态估计[J].华侨大学学报(自然科学版),2023,44(6):751-758.[doi:10.11830/ISSN.1000-5013.202305020]
　LIN Zhihong,ZHENG Lixin,ZENG Yuanyue.Pose Estimation of Escalator Dangerous Behavior Using Spatially-Aware Multi-Task Decoupled Pose Network[J].Journal of Huaqiao University(Natural Science),2023,44(6):751-758.[doi:10.11830/ISSN.1000-5013.202305020]
点击复制

采用空间依赖的MTDPN扶梯危险行为的姿态估计()
分享到：

《华侨大学学报（自然科学版）》[ISSN:1000-5013/CN:35-1079/N]

卷:
第44卷
期数:
2023年第6期
页码:
751-758
栏目:
出版日期:
2023-11-20

文章信息/Info

Title:
Pose Estimation of Escalator Dangerous Behavior Using Spatially-Aware Multi-Task Decoupled Pose Network
文章编号:
1000-5013(2023)06-0751-08
作者:
林志鸿1;  郑力新1;  曾远跃2
1. 华侨大学 工学院, 福建 泉州 362021;2. 福建省特种设备检验研究院 泉州分院, 福建 泉州 362021
Author(s):
LIN Zhihong1;  ZHENG Lixin1;  ZENG Yuanyue2
1. College of Engineering, Huaqiao University, Quanzhou 362021, China; 2. Quanzhou Branch, Fujian Special Equipment Inspection and Research Institute, Quanzhou 362021, China
关键词:
自动扶梯;  人体姿态估计;  危险行为检测;  任务解耦;  空间依赖
Keywords:
escalator;  human pose estimation;  dangerous behavior detection;  task decoupling;  spatially-aware
分类号:
TP391.41;TU229
DOI:
10.11830/ISSN.1000-5013.202305020
文献标志码:
A
摘要:
为实现自动扶梯场景下姿态估计的快速响应和准确估计,提出一种基于空间依赖的多任务解耦姿态网络(MTDPN)。首先,对姿态估计网络进行定位和分类任务分支的解耦,使每个任务分支能够自适应地调整特征关注方向;其次,提出一种空间依赖卷积,通道联合层和空间联合层作为中间层,以逐点卷积和逐深度卷积取代传统卷积,从而降低MTDPN的参数量和浮点计算量,使每张图片的检测时间仅为73.3 ms。在扶梯危险行为关键点数据集和COCO关键点数据集上对MTDPN进行评估。结果表明:与原始网络YOLOPOSE相比,MTDPN在扶梯危险行为关键点数据集和COCO关键点数据集上的准确性指标均有所提高。
Abstract:
In order to realize fast response and accurate estimation of pose estimation in escalator scenarios, a multi-task decoupled pose network(MTDPN)based on spatially-aware is proposed. Firstly, the localization and classification task branches are decoupled for the pose estimation network so that each task branch can adaptively adjust the feature focus direction. Secondly, a spatially-aware convolution is proposed, with the channel joint layer and the spatial joint layer as the intermediate layer, replacing traditional convolution with point wise convolution and depth wise convolution, thus reducing the number of parameters and the computation of floating point of the MTDPN, so that the detection time of each image is only 73.3 ms. The MTDPN is evaluated on the escalator dangerous behavior key point dataset and the COCO key point dataset. The results show that the MTDPN has improved accuracy metrics on both the escalator dangerous behavior key point dataset and COCO key point dataset compared to the original network YOLOPOSE.

参考文献/References:

[1] 舒文华,欧阳惠卿.自动扶梯乘客行为智能感知和自主安全管理技术标准探讨[J].质量与标准化,2021,354(10):39-43.DOI:10.3969/j.issn.2095-0918.2021.10.015.
[2] 蒋儒浩.自动扶梯综合性能检测仪研制[D].合肥:合肥工业大学,2019.
[3] 付春平.自动扶梯几起安全事故的共性分析与探讨[J].科技与创新,2023,217(1):82-84,89.DOI:10.15913/j.cnki.kjycx.2023.01.023.
[4] 张栓柱.基于事故树的商场电梯事故分析[J].消防界(电子版),2022,8(21):21-23.DOI:10.16859/j.cnki.cn12-9204/tu.2022.21.040.
[5] 解云蕾.自动扶梯安全探讨[J].中国科技信息,2022(3):64-66.DOI:10.3969/j.issn.1001-8972.2022.03.021.
[6] CHEN Yucheng,TIAN Yingli,HE Mingyi.Monocular human pose estimation: A survey of deep learning-based methods[J].Computer Vision and Image Understanding,2020,192:102897.DOI:10.1016/j.cviu.2019.102897.
[7] ZHENG Ce,WU Wenhan,CHEN Chen,et al.Deep learning-based human pose estimation: A survey[EB/OL].(2023-07-03)[2023-09-19] .https://doi.org/10.48550/arXiv.2012.13392.
[8] 汤一平,杨冠宝,胡飞虎,等.基于计算机视觉的自动扶梯节能系统[J].计算机测量与控制,2011,19(7):1659-1661,1677.DOI:10.16526/j.cnki.11-4762/tp.2011.07.052.
[9] CAO Zhe,HIDALGO G,SIMON T,et al.OpenPose: Realtime multi-person 2D pose estimation using part affinity fields[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2021,43(1):172-186.DOI:10.1109/TPAMI.2019.2929257.
[10] CHENG B,XIAO Bin,WANG Jingdong,et al.HigherHRNet: Scale-aware representation learning for bottom-up human pose estimation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE Press,2020:5386-5395.DOI:10.48550/arXiv.1908.10357.
[11] NEFF C,SHETH A,FURGURSON S,et al.Efficienthrnet: Efficient scaling for lightweight high-resolution multi-person pose estimation[EB/OL].(2020-12-30)[2023-09-19] .https://doi.org/10.48550/arXiv.2007.08090.
[12] 田联房,吴啟超,杜启亮,等.基于人体骨架序列的手扶电梯乘客异常行为识别[J].华南理工大学学报(自然科学版),2019,47(4):10-19.DOI:10.12141/j.issn.1000-565X.180186.
[13] FANG Haoshu,XIE Shuqin,LU Cewu,et al.RMPE: Regional multi-person pose estimation[C]//Proceedings of the IEEE International Conference on Computer Vision.Venice:IEEE Press,2017:2334-2343.DOI:10.48550/arXiv.1612.00137.
[14] SUN Ke,XIAO Bin,LIU Dong,et al.Deep high-resolution representation learning for human pose estimation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach:IEEE Press,2019:5693-5703.DOI:10.1109/CVPR.2019.00584.
[15] CHEN Yilun,WANG Zhicheng,PENG Yuxiang,et al.Cascaded pyramid network for multi-person pose estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE Press,2018:7103-7112.DOI:10.48550/arXiv.1711.07319.
[16] ASGARY S,MOTAZEDIAN H R,PARIROKH M,et al.KAPAO: A MEMS-based natural guide star adaptive optics system[J].Iranian Endodontic Journal,2013,8(1):1-5.DOI:10.1117/12.2005959.
[17] PAPANDREOU G,ZHU T,KANAZAWA N,et al.Towards accurate multi-person pose estimation in the wild[C]//IEEE Conference on Computer Vision and Pattern Recognition.Honolulu:IEEE Press,2017:3711-3719.DOI:10.1109/CVPR.2017.395.
[18] 杨学存,李杰华,陈丽媛,等.基于人体骨架的扶梯乘客异常行为识别方法[J/OL].安全与环境学报,2022:1-9[2023-06-25] .DOI:10.13637/j.issn.1009-6094.2022.2404.
[19] VASWANI A,SHAZEER N,PARMAR N,et al.Attention is all you need[J].Advances in Neural Information Processing Systems,2017,30:6000-6010.DOI:10.48550/arXiv.1706.03762.
[20] LI Yanjie,ZHANG Shoukui,WANG Zhicheng,et al.TokenPose: Learning keypoint tokens for human pose estimation[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.Montreal:IEEE Press,2021:11313-11322.DOI:10.48550/arXiv.2104.03516.
[21] LI Ke,WANG Shijie,ZHANG Xiang,et al.Pose recognition with cascade transformers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville:IEEE Press,2021:1944-1953.DOI:10.1109/CVPR46437.2021.00198.
[22] MAJI D,NAGORI S,MATHEW M,et al.YOLO-Pose: Enhancing YOLO for multi person pose estimation using object keypoint similarity loss[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans:IEEE Press,2022:2637-2646.DOI:10.48550/arXiv.2204.06806.
[23] REDMON J,FARHADI A.YOLOv3: An incremental improvement[C]//Computer Vision and Pattern Recognition.Berlin:Springer,2018,1804:1-6.DOI:10.48550/arXiv.1804.02767.
[24] BOCHKOVSKIY A,WANG C Y,LIAO H Y M.YOLOv4: Optimal speed and accuracy of object detection[EB/OL].(2020-04-23)[2023-09-08] .https://doi.org/10.48550/arXiv.2004.10934.
[25] WU Yue,CHEN Yinpeng,YUAN Lu,et al.Rethinking classification and localization for object detection[EB/OL].(2020-04-02)[2023-09-07] .https://doi.org/10.48550/arXiv.1904.06493.
[26] SONG Guanglu,LIU Yu,WANG Xiaogang.Revisiting the sibling head in object detector[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE Press,2020:11563-11572.DOI:10.1109/CVPR42600.2020.01158.
[27] MA Ningning,ZHANG Xiangyu,ZHENG Haitao.ShuffleNet V2: Practical guidelines for efficient CNN architecture design[C]//European Conference on Computer Vision.[S.l.]:Springer,2018:122-138.DOI:10.1007/978-3-030-01264-9_8.
[28] TAN Mingxing,LE Q V.EfficientNetV2: Smaller models and faster training[C]//International Conference on Machine Learning.[S.l.]:PMLR,2021:10096-10106.DOI:10.48550/arXiv.2104.00298.
[29] WANG C Y,BOCHKOVSKIY A,LIAO H Y M.YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition.Vancouver:IEEE Press,2023:7464-7475.DOI:10.1109/CVPR52729.2023.00721.
[30] NEWELL A,YANG Kaiyu,DENG Jia.Stacked hourglass networks for human pose estimation[EB/OL].(2016-07-26)[2023-09-07] .https://doi.org/10.48550/arXiv.1603.06937.
[31] KREISS S,BERTONI L,ALAHI A.PifPaf: Composite fields for human pose estimation[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach:IEEE Press,2019:11969-11978.DOI:10.1109/CVPR.2019.01225.
[32] CAO Xianshuai,SHI Yuliang,YU Han,et al.DEKR: Description enhanced knowledge graph for machine learning method recommendation[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information.[S.l.]:ACM,2021:203-212.DOI:10.1145/3404835.3462900.

备注/Memo

备注/Memo:
收稿日期: 2023-05-29
通信作者: 郑力新(1967-),男,博士,教授,主要从事图像分析、机器视觉和深度学习方法的研究。E-mail:zlx@hqu.edu.cn。
基金项目: 福建省科技计划项目(2020Y0039); 福建省泉州市科技计划项目(2020C042R)http://www.hdxb.hqu.edu.cn
更新日期/Last Update: 2023-11-20