2024
doi: 10.12720/jait.15.12.1374-1379
Automatic Segmentation of Table Tennis Match Video Clips Based on Player Actions for Enhanced Data Acquisition
- Department of Electronic Engineering, National Ilan University, Ilan County, Taiwan
- Manuscript receivedJuly 10, 2024
- revisedAugust 10, 2024
- acceptedSeptember 9, 2024
- publishedDecember 12, 2024
Abstract
The collection of data in table tennis competitions is crucial for professional coaches and top players. Big data provides them with insights to analyze opponents’ strategies and adjust their own training directions. However, existing models for recognizing table tennis serving and receiving actions still suffer from low accuracy, making them unsuitable for automated match data recording. Currently, training data for machine learning models are sourced from manually edited videos with labels, which is time-consuming, requires professionals, and generates limited data, leading to poor accuracy in existing action recognition models. Therefore, this study aims to achieve automatic video clipping of table tennis matches using computer vision and machine learning techniques to enhance data collection efficiency. This innovative approach reduces the need for human resources, speeds up data generation, and ensures high accuracy in segment identification. This research effectively automates the labor-intensive process of data preparation, solving the problem of obtaining large, annotated datasets for training machine learning models. Furthermore, this automated clipping framework can be applied to other action recognition tasks, improving the accuracy of semi-supervised learning models in various domains. The proposed Table Tennis Playing Status Recognition Model (TTPSRM) achieved a serve recognition accuracy of 98.4% and a receive recognition accuracy of 96.6%. The automated video segmentation system, tested with 2022 International Table Tennis Federation (ITTF) competition videos, successfully clipped 70% of serve actions, with average start and end point errors of 0.547 and 0.287 s, respectively.
Keywords
- table tennis
- Gated Recurrent Unit (GRU)
- human pose estimation
- OpenPose
- automatic clipping
How to Cite
Zhong-Kai Wei and Jieh-Ren Chang, "Automatic Segmentation of Table Tennis Match Video Clips Based on Player Actions for Enhanced Data Acquisition," Journal of Advances in Information Technology, vol. 15, no. 12, pp. 1374-1379, 2024. https://doi.org/10.12720/jait.15.12.1374-1379
Copyright & License
Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).






