SportPoseNet: Leveraging Pose Estimation and Deep Learning for Sports Activity Classification

Saakshi M V, Dhanya Rao, Nidhi M, Anurag Hurkadli, Sneha Varur

2025

Abstract

SportPoseNet is designed to recognize and classify sports activities by combining advanced pose estimation techniques with the powerful ResNet-200 architecture. To support this effort, we curated a custom sports dataset tailored specifically for training and evaluation purposes. By utilizing MediaPipe, the system accurately extracts keypoints to create pose landmarks that represent a wide range of sports activities.These pose landmarks are then processed by a fine-tuned ResNet-200 model, which achieves an impressive validation accuracy of 92.67. The system is capable of classifying activities like cricket, badminton, hockey, and football with remarkable precision. By blending the lightweight and efficient pose estimation capabilities of Medi-aPipe with the robust performance of ResNet-200, SportPoseNet provides a scalable and accurate solution for real-time sports activity recognition. This approach not only demonstrates the power of custom datasets and cutting-edge technologies in sports analysis but also paves the way for improved athlete monitoring and performance insights.

Paper Citation

in Harvard Style

M V S., Rao D., M N., Hurkadli A. and Varur S. (2025). SportPoseNet: Leveraging Pose Estimation and Deep Learning for Sports Activity Classification. In Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT; ISBN 978-989-758-763-4, SciTePress, pages 581-588. DOI: 10.5220/0013632600004664

in Bibtex Style

@conference{incoft25,
author={Saakshi M V and Dhanya Rao and Nidhi M and Anurag Hurkadli and Sneha Varur},
title={SportPoseNet: Leveraging Pose Estimation and Deep Learning for Sports Activity Classification},
booktitle={Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT},
year={2025},
pages={581-588},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013632600004664},
isbn={978-989-758-763-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT
TI - SportPoseNet: Leveraging Pose Estimation and Deep Learning for Sports Activity Classification
SN - 978-989-758-763-4
AU - M V S.
AU - Rao D.
AU - M N.
AU - Hurkadli A.
AU - Varur S.
PY - 2025
SP - 581
EP - 588
DO - 10.5220/0013632600004664
PB - SciTePress