Performance evaluation of deep learning models for sports image classification

With the advancement of computer vision, sports image classification has become essential in areas such as sports event analysis and athlete motion recognition. Deep learning, particularly Convolutional Neural Networks (CNNs), has shown outstanding performance in these tasks. This study compares three widely used deep learning models—MobileNet, ResNet50, and VGG16—to evaluate their effectiveness in classifying sports images. A dedicated sports image dataset was used for training and testing. The models were assessed based on training loss, validation loss, and accuracy curves. Results show that MobileNet and ResNet50 converge faster and yield lower validation loss than VGG16. These models are better suited for real-time applications and environments with limited computational resources.