一种基于改进3D卷积的乒乓球球员动作识别算法
A Recognition Algorithm of Player Motion of Table Tennis Based on Improved 3D Convolution
DOI: 10.12677/CSA.2021.116185, PDF,    科研立项经费支持
作者: 汪语哲：大连民族大学大数据应用技术国家民委重点实验室，辽宁 大连;大连民族大学大连市民族文化数字技术重点实验室，辽宁 大连;大连民族大学机电工程学院，辽宁 大连；刘明方, 尹真杰, 段晓东：大连民族大学大数据应用技术国家民委重点实验室，辽宁 大连;大连民族大学大连市民族文化数字技术重点实验室，辽宁 大连;大连民族大学计算机科学与工程学院，辽宁 大连
关键词: 乒乓球击球；人体动作识别；3D卷积网络；动作分类；视频跟踪；Striking of Table Tennis； Human Action Recognition； 3D Convolutional Network； Action Classification； Video Tracking
摘要: 文章研究了基于视频的乒乓球球员动作识别问题。在计算机视觉领域，人体动作识别具有一定挑战性。基于专业乒乓球运动员在乒乓球发球机的接发动作视频，构建了乒乓球击球动作视频数据集，将其分为正手击球、反手击球、正手拉球、反手拉球和非击球动作5类。提出通过人体密集姿态(Dense Pose)处理数据集，将把人体形态从环境中进行提取，随后提出一种改进的C3D卷积网络，用于学习数据集上连续帧的时空特征。结果表明，文章设计的算法对于光线、环境等干扰因素具有较好的鲁棒性，泛化性能好，为基于视频的动作分类识别问题提出了一种可行解决方案。
Abstract: Motion recognition of table tennis players based on video is studied in this paper. Recognition of human action is challenging in the field of computer vision. Based on videos of ball strike of professional table tennis players against table tennis ball machine, a data set of ball strike of table tennis players is constructed and divided into 5 catalogs of forehand shots, backhand shots, forehand shots, backhand shots and non-stike action. Dense pose of the human body is used to process the constructed data set and extract human body shape from the environment, and then an improved C3D convolutional network is proposed to learn the spatiotemporal features of continuous frames on the data set. Results show that the algorithm proposed in the article has good robustness to interference factors such as light and environment, and good generalization performance, demonstrating a feasible solution to the problem of video-based action classification and recognition.
文章引用：汪语哲, 刘明方, 尹真杰, 段晓东. 一种基于改进3D卷积的乒乓球球员动作识别算法[J]. 计算机科学与应用, 2021, 11(6): 1791-1801. https://doi.org/10.12677/CSA.2021.116185

参考文献

[1] 王恺凡. 基于人脸识别的乒乓球智能训练平台设计[D]: [硕士学位论文]. 南京: 南京邮电大学, 2020.
[2] 丁朔. 基于智能语音交互的乒乓球训练系统的设计与实现[D]: [硕士学位论文]. 南京: 南京邮电大学, 2020.
[3] 杨波. 虚拟现实技术应用于高校乒乓球教学中的实证研究[D]: [硕士学位论文]. 兰州: 西北师范大学, 2020.
[4] 任云青. 智能乒乓球自动捡球机器人的设计与实现[D]: [硕士学位论文]. 南京: 南京邮电大学, 2020.
[5] 孙于成. 基于时空图卷积的乒乓球基础技术动作识别[D]: [硕士学位论文].安庆: 安庆师范大学, 2020.
[6] Martin, P.-E., Benois-Pineau, J., Péteri, R. and Morlier, J. (2020) 3D Attention Mechanism for Fine-Grained classification of table ten-nis strokes using a Twin Spatio-Temporal Convolutional Neural Networks. 25th International Conference on Pattern Recognition, Milano, January 2021. arXiv preprint arXiv:2012.05342.
[7] 杨静. 体育视频中羽毛球运动员的动作识别[J]. 自动化技术与应用, 2018, 37(10): 120-124.
[8] binti Rahmad, N.A., binti Sufri, N.A. J, bin As’ari, M.A., et al. (2019) Recognition of Badminton Action Using Convolutional Neural Network. Indonesian Journal of Electrical En-gineering and Informatics (IJEEI), 7, 750-756. [Google Scholar] [CrossRef] 
[9] Piergiovanni, A.J. and Ryoo, M.S. (2018) Fine-Grained Activity Recognition in Baseball Videos. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, 18-22 June 2018, 1740-1748. [Google Scholar] [CrossRef] 
[10] Tran, D., Bourdev, L., Fergus, R., et al. (2015) Learning Spati-otemporal Features with 3D Convolutional Networks. 2015 IEEE International Conference on Computer Vision, San-tiago, 7-13 December 2015, 4489-4497. [Google Scholar] [CrossRef] 
[11] Shao, D., Zhao, Y., Dai, B. and Liu, D. (2020) FineGym: A Hierar-chical Video Dataset for Fine-Grained Action Understanding. 2020 IEEE/CVF Conference on Computer Vision and Pat-tern Recognition (CVPR), 13-19 June 2020, Seattle, 2616-2625. [Google Scholar] [CrossRef] 
[12] Güler, R.A., Neverova, N. and Kokkinos, I. (2018) Densepose: Dense Human Pose Estimation in the Wild. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 18-23 June 2018, 7297-7306. [Google Scholar] [CrossRef] 
[13] Neverova, N., Guler, R.A. and Kokkinos, I. (2018) Dense Pose Transfer. In: Ferrari, V., Hebert, M., Sminchisescu, C. and Weiss, Y., Eds., Proceedings of the European Conference on Computer Vision (ECCV), Springer, Cham, 128-143. [Google Scholar] [CrossRef] 
[14] 机器之心Pro.Facebook实时人体姿态估计: Dense Pose及其应用展望[EB/OL].
https://baijiahao.baidu.com/s?id=1625055353488715502&wfr=spider&for=pc, 2019-02-01.
[15] LeCun, Y., Bot-tou, L., Bengio, Y. and Haffner, P. (1998) Gradient-Based Learning Applied to Document Recognition. Proceedings of the IEEE, 86, 2278-2324. [Google Scholar] [CrossRef] 
[16] Soomro, K., Zamir, A.R. and Shah, M. (2012) UCF101: A Dataset of 101 Human Actions Classes from Videos in the Wild. CoRR, 1212, 0402.