Artificial Intelligence (AI) virtual coaches are currently a research hotspot in the field of sports and health. To improve the effectiveness of fitness training, this article proposes an intelligent virtual coaching system based on reinforcement learning and motion capture technology. Specifically, by collecting user motion data and utilizing two-dimensional human pose estimation methods for motion capture technology, a training model is constructed. Secondly, a reinforcement learning reward function combining imitation rewards and adaptive penalties was designed to balance the needs of action imitation and diversity adaptation, thereby generating a natural and stable training action sequence. Then, to alleviate the training difficulty caused by the initial posture, a dynamic weight adjustment mechanism was proposed. Finally, the system was experimentally validated on COCO, MPII, and self-built datasets, and the experimental results showed that the system performed excellently in terms of action accuracy, adaptability, and user satisfaction.
StankovskaGMemediIDimitrovskiD. Coronavirus COVID-19 disease, mental health and psychosocial support. Society Register2020; 4(2): 33–48.
2.
ChenZ. Artificial intelligence-virtual trainer: Innovative didactics aimed at personalized training needs. J Knowl Econ2023; 14(2): 2007–2025.
3.
SalemiSHamidiMMoghadasY, et al.Strategy of preventing sports injuries of students in schools with grounded theory approach. Physical Treatments-Specific Physical Therapy Journal2023; 13(4): 287–294.
4.
GlowinskiSMajdanikS. Fatal consequences of safety non-compliance in non-commercial ultralight aviation: a case study. Saf Now2024; 10(3): 77.
5.
BaumgartnerTKlattS. Monocular 3d human pose estimation for sports broadcasts using partial sports field registration. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. Piscataway: IEEE, 2023, pp. 5109–5118.
6.
ChenDSongYLiangF, et al.3D human body reconstruction based on SMPL model. Vis Comput2023; 39(5): 1893–1906.
7.
JiangCXiaoYWuC, et al.A2j-transformer: anchor-to-joint transformer network for 3d interacting hand pose estimation from a single rgb image. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023, pp. 8846–8855.
8.
WangGZengHWangZ, et al.Motion projection consistency-based 3-D human pose estimation with virtual bones from monocular videos. IEEE Trans Cogn Dev Syst2022; 15(2): 784–793.
9.
XuJYuZNiB, et al.Deep kinematics analysis for monocular 3d human pose estimation. In: Proceedings of the IEEE/CVF Conference on computer vision and Pattern recognition. Piscataway: IEEE, 2020, pp. 899–908.
10.
LinJLiSQinH, et al.Overview of 3d human pose estimation. Comput Model Eng Sci2023; 134(3): 1621–1651.
11.
LeeKLeeILeeS. Propagating lstm: 3d pose estimation based on joint interdependency. In: Proceedings of the European conference on computer vision (ECCV). Berlin, Germany: Springer, 2018, pp. 119–135.
12.
XuJYuZNiB, et al.Deep kinematics analysis for monocular 3d human pose estimation. In: Proceedings of the IEEE/CVF Conference on computer vision and Pattern recognition. Piscataway: IEEE, 2020, pp. 899–908.
13.
ZhengCWuWChenC, et al.Deep learning-based human pose estimation: a survey. ACM Comput Surv2023; 56(1): 1–37.
14.
LinJLiSQinH, et al.Overview of 3d human pose estimation. Comput Model Eng Sci2023; 134(3): 1621–1651.
15.
LiSChanAB. 3d human pose estimation from monocular images with deep convolutional neural network. In: Computer Vision--ACCV 2014: 12th Asian Conference on Computer Vision, Singapore, Singapore, November 1-5, 2014, Revised Selected Papers, Part II 12. Berlin, Germany: Springer International Publishing, 2015, pp. 332–347.
16.
ZhouLMengXLiuZ, et al.Human pose-based estimation, tracking and action recognition with deep learning: a survey. arXiv preprint arXiv:2310.130392023.
17.
HeJYangW. Video-based human pose regression via decoupled space-time aggregation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2024, pp. 1022–1031.
18.
DuSWangHYuanZ, et al.Bi-pose: bidirectional 2D-3D transformation for human pose estimation from a monocular camera. IEEE Trans Autom Sci Eng2023.
19.
ZhouKHanXJiangN, et al.Hemlets pose: learning part-centric heatmap triplets for accurate 3d human pose estimation. In: Proceedings of the IEEE/CVF international conference on computer vision. Piscataway: IEEE, 2019, pp. 2344–2353.
20.
WangMChenXLiuW, et al. (2018). Drpose3d: depth ranking in 3d human pose estimation. arXiv preprint arXiv:1805.08973.
21.
AndrilukaMPishchulinLGehlerP, et al.2d human pose estimation: new benchmark and state of the art analysis. In: Proceedings of the IEEE Conference on computer Vision and Pattern Recognition. Piscataway: IEEE, 2014, pp. 3686–3693.
22.
LinTYMaireMBelongieS, et al.Microsoft coco: common objects in context. In: Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13. Berlin, Germany: Springer International Publishing, 2014, pp. 740–755.
23.
HwangDHKimSMonetN, et al.Lightweight 3d human pose estimation network training using teacher-student learning. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. Piscataway: IEEE, 2020, pp. 479–488.
24.
ChenZHuangYYuH, et al.Towards part-aware monocular 3d human pose estimation: an architecture search approach. In: Computer vision–ECCV 2020: 16th European conference, Glasgow, UK, August 23–28, 2020, proceedings, Part III. Berlin, Germany: Springer International Publishing, 2020, vol 16, pp. 715–732.
25.
InsafutdinovEPishchulinLAndresB, et al.Deepercut: a deeper, stronger, and faster multi-person pose estimation model. In: Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part VI 14. Berlin, Germany: Springer International Publishing, 2016, pp. 34–50.
