Basketball Activity Classification Based on Upper Body Kinematics and Dynamic Time Warping

Area of Science:

• Sports Science
• Biomechanics
• Machine Learning

Background:

• Traditional basketball activity classification methods (manual, multi-camera) have limitations in accuracy and spatial constraints.
• Accurate player activity classification is crucial for performance analysis, training optimization, and injury prevention.

Purpose of the Study:

• To develop and evaluate a novel basketball activity classification model using Dynamic Time Warping (DTW) and body kinematic measures.
• To assess the model's accuracy and efficiency in classifying four fundamental basketball movements: shooting, passing, dribbling, and lay-up.

Main Methods:

• An experimental study involving 20 participants (10 experienced, 10 novice players).
• Classification of four basketball movements based on kinematic measures, primarily resultant hand acceleration.
• Application of Dynamic Time Warping (DTW) algorithm for movement pattern recognition.

Main Results:

• The proposed DTW-based model achieved high accuracy in classifying basketball movements.
• Classification precision, recall, and specificity reached up to 0.984, 0.983, and 0.994, respectively, using resultant hand acceleration.
• The model demonstrated efficiency for real-time sports activity classification.

Conclusions:

• Dynamic Time Warping (DTW) is feasible for real-time basketball activity classification.
• The study provides insights into optimal sensor placement for accurate classification.
• This approach supports quantitative supervision of player exertion and training strategy optimization.