Abstract

Recognizing various types of sports fields, such as swimming pools, basketball courts, and tennis courts, can be challenging due to their differing shapes, sizes, colors, and orientations worldwide. In this article, a novel approach to sports field classification is proposed using a dataset of approximately 7700 images. The classification task utilizes convolutional neural network (CNN) algorithms, which are widely employed in image recognition tasks as a deep learning technique. The results demonstrate that CNN-driven sports field classification applications can effectively automate the process of identifying different types of sports fields. This not only aids in easier identification of different types of sports field locations but also opens new avenues for sports field-related infrastructure planning and analysis. The trained model achieved an accuracy of 92.71% on the test set, validating the practicality and effectiveness of this approach