基于增强高分辨率网络的乒乓球击球者姿态估计
Human Pose Estimation of Table Tennis Player Based on Enhanced High Resolution Network
DOI: 10.12677/CSA.2022.1212271, PDF,   
作者: 申雨钰, 王直杰：东华大学，信息科学与技术学院，上海
关键词: 人体姿态估计；高分辨率网络；YOLOv5；Ghost模块；Human Pose Estimation； HRNet； YOLOv5； Ghost Module
摘要: 乒乓球旋转球的识别是乒乓球机器人视觉系统设计中亟待解决的难题，而对乒乓球旋转变化的识别离不开对击球者挥拍姿态的研究，因此本文针对乒乓球击球者姿态分析提出了一种基于增强高分辨率网络的姿态估计算法。该算法融入Ghost模块的YOLOv5人体检测模型结合高分辨率网络姿态估计模型，减少模型的参数量，提升模型运算速度。最后在本文自制的数据集PP-Person上的实验结果表明，本文提出的姿态估计算法有效降低了网络参数量，在保持一定预测精度的情况下，响应速度较HRNet提高了55.76%。
Abstract: The recognition of table tennis rotation ball is a difficult problem to be solved in the design of table tennis robot vision system, and the recognition of table tennis rotation changes can not be separated from the research on the player’s swing posture. Therefore, this paper proposes a pose estimation algorithm based on enhanced high resolution network for the analysis of table tennis player’s posture. The algorithm integrates YOLOv5 human detection model of Ghost module with high resolution network human pose estimation model, reducing the model parameters and improving the model operation speed. Finally, the experimental results on the self-made data set PP-Person in this paper show that the estimation algorithm proposed in this paper effectively reduces the number of network parameters, and with a certain prediction accuracy, the response speed is increased by 55.76% compared with HRNet.
文章引用：申雨钰, 王直杰. 基于增强高分辨率网络的乒乓球击球者姿态估计[J]. 计算机科学与应用, 2022, 12(12): 2675-2683. https://doi.org/10.12677/CSA.2022.1212271

参考文献

[1] 章逸丰. 快速飞行物体的状态估计和轨迹预测[D]: [博士学位论文]. 杭州: 浙江大学, 2015: 1-7.
[2] 赵永生. 旋转飞行乒乓球的状态估计和轨迹预测[D]: [博士学位论文]. 杭州: 浙江大学, 2017: 29-56.
[3] 任艳青, 方灶军, 徐德, 谭民. 基于模糊神经网络的乒乓球旋转飞行轨迹模式分类[J]. 控制与决策, 2014, 29(2): 263-269.
[4] Wang, H., Cao, C. and Leung, H. (2006) An Improved Locally Weighted Regression for a Converter Re-Vanadium Prediction Modeling. 2006 6th World Congress on Intelligent Control and Automation, Dalian, 21-23 June 2006, 1515-1519.
[5] Huang, Y., Xu, D., Tan, M. and Su, H. (2011) Trajectory Prediction of Spinning Ball for Ping-Pong Player Robot. 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco, 25-30 September 2011, 3434-3439. [Google Scholar] [CrossRef] 
[6] 谈小峰. 人体检测和姿态识别在乒乓球机器人上的应用研究[D]: [硕士学位论文]. 上海: 东华大学, 2021: 17-72.
[7] Gao, Y., Tebbe, J., Krismer, J. and Zell, A. (2019) Markerless Racket Pose Detection and Stroke Classification Based on Stereo Vision for Table Tennis Robots. 2019 3rd IEEE International Conference on Robotic Computing, Naples, 25-27 February 2019, 189-196. [Google Scholar] [CrossRef] 
[8] 索芳菲, 季云峰. 乒乓球机器人核心算法研究综述[J]. 电子科技, 2022: 1-8.
[9] Cao Z, Simon, T., Wei, S.-E. and Sheikh, Y. (2017) Realtime Multi-Person 2D Pose Esti-mation Using Part Affinity Field. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Hono-lulu, 21-26 July 2017, 1302-1310. [Google Scholar] [CrossRef] 
[10] Toshev, A. and Szegedy, C. (2014) DeepPose: Human Pose Estimation via Deep Neural Networks. 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, 23-28 June 2014, 1653-1660. [Google Scholar] [CrossRef] 
[11] Wei, S.-E., Ramakrishna, V., Kanade, T. and Sheikh, Y. (2016) Convolutional Pose Machines. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 27-30 June 2016, 4724-4732. [Google Scholar] [CrossRef] 
[12] Newell, A., Yang, K. and Deng, J. (2016) Stacked Hourglass Net-works for Human Pose Estimation. In: Leibe, B., Matas, J., Sebe, N. and Welling, M., Eds., Computer Vision—ECCV 2016, Springer, Cham, 483-499. [Google Scholar] [CrossRef] 
[13] Sun, K.，Xiao, B., Liu, D. and Wang, J. (2019) Deep High-Resolution Representation Learning for Human Pose Estimation. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, 15-20 June 2019, 5686-5696. [Google Scholar] [CrossRef] 
[14] Cheng, B., Xiao, B., Wang, J., et al. (2020) HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation. 2020 IEEE/CVF Conference on Comput-er Vision and Pattern Recognition, Seattle, 13-19 June 2020, 5385-5394. [Google Scholar] [CrossRef] 
[15] 钟宝荣, 吴夏灵. 基于高分辨率网络的轻量型人体姿态估计研究[J]. 计算机工程, 2022: 1-10.