Enhancing Movement Quality Evaluation in Sports Education Using Temporal Action Scoring Network: A Deep Learning Approach for Real-time Feedback

Accurately assessing movement quality in sports and physical education is crucial for improving performance. Traditional methods often rely on subjective judgment and fail to provide real-time feedback, limiting their effectiveness in real-world training scenarios. This study proposes the Temporal Action Scoring Network (TASN), a deep learning-based model that integrates Convolutional Neural Networks (CNNs) for spatial feature extraction and Long Short-Term Memory (LSTM) units for temporal sequence modeling. TASN processes motion data captured through wearable sensors or motion capture systems, enabling the extraction of spatial and temporal features to assess movement quality dynamically. Key experimental results demonstrate that TASN outperforms existing state-of-the-art models, achieving lower Mean Squared Error (MSE), higher Pearson Correlation, and a low processing latency of 150 ms. Additionally, an ablation study confirms the critical role of CNN and LSTM components in improving model accuracy. By providing real-time, objective feedback, TASN enhances the precision of movement evaluation and offers actionable insights for athletes and educators to refine techniques and optimize performance. This research establishes a scalable, efficient framework for movement quality assessment with applications in sports training and physical education.