Design and application of a handball training level evaluation system for college students.

This article focuses on the design and development of a comprehensive handball training evaluation system for college students using the Azure Kinect dynamic body sensing device combined with the Attentional Factorization Machine (AFM) algorithm. The system captures three-dimensional skeletal motion data in real time and quantitatively assesses athletes' movements by comparing them with a standardized database, revealing detailed differences between college players and professional athletes in key actions such as defense stances, passing, and shooting. Analysis identified slow ball speed as the primary factor affecting shooting success, and the study highlights the system's potential to provide precise, data-driven feedback to improve training quality and athlete performance. Limitations include a relatively small and homogeneous sample and unexamined environmental factors, suggesting future research should broaden participant diversity and control for external variables to enhance applicability.