Mathematics and Computational Sciences

A computational framework for sports analytics: Player tracking and strider rate estimation using deep learning

Document Type : Original Article

Authors

Mahatishri Jayakumar 1
Prabu Kanth 2
Kevin John Chitralekha 1
Sam Varghese George 2

1 Department of Artificial Intelligence and Data Science, St. Joseph’s College Of Engineering, Chennai, India.

2 Department of Electronics and Communication Engineering, St. Joseph’s College Of Engineering, Chennai, India.

https://doi.org/10.30511/mcs.2026.2077776.1620
Abstract
This paper presents a computational framework for estimating stride rate in football using advanced computer vision and deep learning techniques, integrating modules for player detection, tracking, team identification, pitch mapping, and performance analysis. Convolutional Neural Networks (CNNs) are used for spatial feature extraction, while YOLOv8 enables accurate player detection, and a Kalman filter supports robust multi-object tracking by modeling player motion as continuous trajectories in two-dimensional Euclidean space. Player kinematics are derived by computing velocity as the time derivative of position and total distance as the cumulative displacement between frames. The system incorporates AlphaPose for anatomical keypoint detection, allowing precise motion capture, and models periodic movement using sinusoidal functions to estimate stride frequency. To enhance accuracy, the Savitzky–Golay filter is applied for trajectory smoothing. Experimental evaluation on broadcast football footage demonstrates strong performance, achieving 93.1% consistency in stride rate estimation, a 90.3% success rate, and an error margin below 2%. Additionally, the integration of digital twinning technology enables real-time visualization of player movements, supporting applications in performance optimization, fatigue monitoring, and injury prevention, thereby advancing automated sports analytics through data-driven decision-making.

Keywords

Stride Rate Estimation
Kalman Filter
Convolutional Neural Networks
Biomechanical Gait Modelling
Euclidean Motion Trajectory
Subjects
[1] Adrian, E. D., & May, R. M. (1927). The dynamics of muscle action. Philosophical Transactions of the Royal Society of London, 215, 339–382.
[2] Barkhordari Firozabadi, S., Shahzadeh Fazeli, S. A., Zarepour Ahmadabadi, J., Karbassi, S. M. S. (2025). Efficient cluster center optimization: A novel hybrid metaheuristic. Mathematics and Computational Sciences, 6(1), 116-146. 
[3] Barrett, S., Ward, P., & Toms, D. (2018). Speed and distance metrics in football player performance analysis. Journal of Sports Science, 36(12), 1399–1407.
[4] Bekrani, M., Zayyani, H. (2025). Affine projection LMS adaptive algorithm with variable smoothing of weight update matrix. Mathematics and Computational Sciences, 6(1), 89-103. 
[5] Bewley, A., Ge, Z., Ott, L., Ramos, F., & Upcroft, B. (2016). Simple online and realtime tracking. In Proceedings of IEEE ICIP (pp. 3464–3468). 
[6] Dapretto, M., Lee, P., & Tsai, S. (2021). Advancing accessibility in sports: Enabling visually impaired athletes through digital twinning. Sports Technology and Accessibility Journal, 3(1), 45–58.
[7] Felzenszwalb, P. F., & Huttenlocher, D. P. (2005). Pictorial structures for object recognition. International Journal of Computer Vision, 61(1), 55–79.
[8] Geert, P., Sam, H., & Ron, B. (2017). Biomechanics of stride rate and efficiency among professional athletes. Journal of Sports Biomechanics, 16(3), 302–315. 
[9] George, S. V., Samyuktha, V., Sanjay, U. (2025, September). Real-Time Multi-Class Car Parking Detection With Imporved Fps Using Transfer Learning Based Instance Segmentation. In 2025 IEEE 4th International Conference for Advancement in Technology (ICONAT) (pp. 1-5). IEEE.
[10] Jabbari, M., Amini, M., Malekinezhad, H., Berahmand, Z. (2023). Improving augmented reality with the help of deep learning methods in the tourism industry. Mathematics and Computational Sciences, 4(2), 33-45.
[11] Johansson, G. (1973). Perception of biological motion. Perception & Psychophysics, 14, 201–211.
[12] Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. In Proceedings of NIPS (pp. 1097–1105).
[13] Laptev, I., Marszalek, M., Schmid, C., & Rozenfeld, B. (2008). Learning realistic human actions from movies. In Proceedings of IEEE CVPR (pp. 1–8).
[14] Liu, Z., et al. (2020). Fusion-based motion tracking. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(3), 601–615.
[15] Mahdi, M., Jabbari, M. (2024). Predicting customer churn in the fast-Moving consumer goods segment of the retail industry using deep learning. Mathematics and Computational Sciences, 5(3), 58-79.
[16] Marr, D. (1982). Vision: A Computational Study of Human Visual Representation and Processing. W.H. Freeman.
[17] Muybridge, E. (1901). The Human Figure in Motion. Dover Publications.
[18] Ramanan, D. (2005). Learning to analyze images of articulated bodies. Advances in Neural Information Processing Systems, 18.
[19] Sharma, R., et al. (2022). Hybrid architectures for motion analysis. In Proceedings of ACM SIGKDD (pp. 1456–1465).
[20] Shi, L., Zhang, Y., Cheng, J., & Lu, H. (2019). Skeleton-based action recognition with multi-stream adaptive graph convolutional networks. In Proceedings of IEEE CVPR (pp. 11428–11437). 
[21] Simonyan, K., & Zisserman, A. (2014). Two-stream convolutional networks for action recognition in videos. Advances in Neural Information Processing Systems, 27.
[22] Sundaram, N., Brox, T., & Keutzer, K. (2010). Dense point trajectories by GPU-accelerated large displacement optical flow. In Proceedings of ECCV (pp. 438–451). 
[23] Tran, D., Bourdev, L., Fergus, R., Torresani, L., & Paluri, M. (2015). Learning spatiotemporal features with 3D convolutional networks. In Proceedings of IEEE ICCV (pp. 4489–4497).
[24] Wojke, N., Bewley, A., & Paulus, D. (2017). Simple online and realtime tracking with a deep association metric. In Proceedings of IEEE ICIP (pp. 3645–3649).
[25] Xiao, B., Wu, H., & Wei, Y. (2018). Simple baselines for human pose estimation and tracking. In Proceedings of IEEE ICCV (pp. 466–474).
[26] Zhang, Y., Li, H., & Huang, G. (2020). Digital twins in smart manufacturing: Fundamentals, applications, and trends. International Journal of Advanced Manufacturing Technology, 107, 3119–3134. 
[27] Zhang, Y., & Wang, L. (2020). FairMOT: On the fairness of detection and re-identification in multiple object tracking. In Proceedings of IEEE CVPR.
Mathematics and Computational Sciences
Volume 7, Issue 2
Spring 2026
Pages 175-188

  • Receive Date 14 November 2025
  • Revise Date 01 May 2026
  • Accept Date 08 May 2026

Home

Submit Manuscript

Reviewers

Contact Us