A Comparative Study of the Application and Performance of Deep Learning in Athlete Performance Evaluation in Sports

Deep learning technology has shown promising applications in athlete performance evaluation and prediction. Researchers have constructed high-quality datasets of athlete performance and designed and trained various deep learning models, including CNNs, LSTMs, attention-enhanced LSTMs, and GCNs. Experimental results demonstrate that the attention-enhanced LSTM model achieves optimal performance in predicting athlete performance, demonstrating the effectiveness of deep learning methods in mining potential patterns and features from time-series data. This study provides new insights and tools for athlete performance evaluation, contributing to the realization of scientific, data-driven sports training and competition management. Future work will focus on optimizing model structures, integrating domain knowledge, and expanding application scenarios, aiming to provide strong support for the development of the sports industry.