Comparison of Four Convolutional Neural Network-Based Algorithms for Sports Image Classification

DOI

10.2991/978-94-6463-370-2_20How to use a DOI?

Keywords

Neural Network; Sports Image Classification; Deep Learning

Abstract

Sports image classification is a challenging task that involves multiple types of sports, with difficulties in feature recognition and suboptimal detection results. This study employs four pretrained models, namely Residual Network 50 (ResNet-50), EfficientNet B7, Densely Connected Convolutional Network 121 (DenseNet-121), and You Only Look Once version 8 (YOLOv8), to address the problem of classifying 100 different sports image categories. The dataset contains 12,200 sports images, which serves as a robust experimental foundation of this research. By comparison their performances it could be found that ResNet-50 exhibited outstanding performance on the training set, achieving an accuracy of 90.80%, and 88.75% on validation set. The EfficientNet B7 model, achieved an accuracy of 37.45% for training and 62.42% for inference. The less impressive performance possibly due to its limited representation capabilities when dealing with specific sports image classification tasks. DenseNet-121 attained an accuracy of 71.791% on the training and 86.211% on the validation set. Compared to EfficientNet B7, its performance is better, suggesting the dense connectivity architecture is more suitable for extracting image features. Furthermore, YOLOv8n model delivered exceptional performance with an average accuracy of 94.90% on the training set, 96.60% on the validation set. These results showcase the robust performance of YOLOv8n in sports image classification and detection. In conclusion, this study provides valuable insights into addressing complex image classification problems by comparing the performance of different algorithms in sports image classification. Understanding the strengths and weaknesses of these various algorithms is crucial for a deeper comprehension of image classification tasks and guiding future research endeavors.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Xiangchen Liu
PY  - 2024
DA  - 2024/02/14
TI  - Comparison of Four Convolutional Neural Network-Based Algorithms for Sports Image Classification
BT  - Proceedings of the 2023 International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2023)
PB  - Atlantis Press
SP  - 178
EP  - 186
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-370-2_20
DO  - 10.2991/978-94-6463-370-2_20
ID  - Liu2024
ER  -