A Proposal of a Spike Event Classification Method Based on Ball Trajectory in a Volleyball Video
DOI:
https://doi.org/10.52731/liir.v004.173Keywords:
Computer Vision, VolleyBall, Sports Analysis, Spike Event ClassificationAbstract
In this research, we propose a method for classifying a spike event in a volleyball game video. In our method, Person and ball detection and tracking are implemented with the YOLO and the ByteTrack. An action event such as a spike and toss is recognized by focusing on the changes in a player’s bounding box of the tracking result. The spike event is classified from the action events with a trajectory of the ball movement. The precision and recall rates of our method are 95% and 93%, respectively, and it indicates an improvement of 8% over existing method.
References
K. Jeroen, H. V.D. Mars, J. V.D. Kamp, W. Walinga, and I. V. Hilvoorde, “Aligning Digital Video Technology with Game Pedagogy in Physical Educa-tion”, J. of Physical Education, Recreation & Dance, Vol. 89, pp. 12–22, Doi. 10.1080/07303084.2017.1390504 (2018)
Z. Weigang, Y. Qixiang, X. Liyuan, and G. Wen, “Unsupervised Sports Video Scene Clustering and Its Applications to Story Units Detection”, Proc. SPIE, Vol. 5960, Doi: 10.1117/12.631389 (2005)
C. C. Hsu, H. T. Chen, C. L. Chou, and S. Y. Lee: “Spiking and Blocking Events Detection and Analysis in Volleyball Videos”, IEEE Int’l Conf. on Multimedia and Expo, pp. 19–24 (2012)
J. Redmon, S. Divvala, R. Girshick, and A. Farhadi: “You Only Look Once: Unified, Real-Time Object Detection”, arXiv CoRR, Vol. abs/1506.02640 (2015)
Z. Yifu, S. Peize, J. Yi, Y. Dongdong, Y. Zehuan, L. Ping, L. Wenyu, and W. Xing-gang: “Multi-Object Tracking by Associating Every Detection Box”, arXiv preprint arXiv:2110.06864 (2021)
