TacPrint: Visualizing the Biomechanical Fingerprint in Table Tennis

Area of Science:

• Sports Science
• Biomechanics
• Machine Learning

Background:

• Table tennis requires high technical skill and coordination.
• Biomechanical fingerprints offer insights into player movement and weaknesses.
• Effective methods for generating biomechanical fingerprints are lacking.

Purpose of the Study:

• To propose TacPrint, a framework for generating biomechanical fingerprints for table tennis players.
• To utilize machine learning and IMU data for feature extraction.
• To enhance model interpretability with attention mechanisms.

Main Methods:

• Developed the TacPrint framework using machine learning.
• Collected biomechanics data via inertial measurement units (IMUs).
• Implemented an attention mechanism for model interpretability and a visualization system for fingerprint exploration.

Main Results:

• The TacPrint framework demonstrated high accuracy and effectiveness in generating biomechanical fingerprints.
• The system successfully facilitated exploration and investigation of player fingerprints.
• Experimental validation confirmed the model's performance.

Conclusions:

• TacPrint provides an effective method for generating biomechanical fingerprints in table tennis.
• The framework can help players identify and improve technical weaknesses.
• TacPrint has potential for application in other sports.