[4Yin-A] Poster 6

[4Yin-A-52] A Study on Skill Level Classification and Biomechanical Validity of Basketball Shooting Motions Using Motion Data

〇Iori Furuya¹, Kazuya Tomabechi², Tatsuki Seino³, Rui Miyazawa⁴ (1. Yokohama City University, 2. The University of Tokyo, 3. Graduate School of Information Science and Technology Hokkaido University, 4. sci-bone Corporation)

Keywords:

Analysis of basketball motion,Biomechanics,Skill classification

The purpose of this study is to examine the consistency between machine-learning-based classification and biomechanical knowledge in basketball shooting motion.
This study classified skill levels using a support vector machine (SVM), with joint-related features estimated from RGB videos via a marker-less pose estimation framework.
Experimental results showed that the average classification accuracy across all joints was 40%.
Furthermore, the analysis of feature contributions revealed that the joints exhibiting high importance in the SVM model tended to correspond to those identified as biomechanically significant in previous studies.
These findings suggest that analyzing the feature contributions of machine-learning-based skill-level classification can provide interpretations consistent with biomechanical knowledge.