Pose estimation for pickleball players' kinematic analysis through MediaPipe-based deep learning: A pilot study

Area of Science:

• Sports Science
• Biomechanics
• Kinesiology

Background:

• Pickleball's rising popularity spans all age groups.
• The sport requires specific techniques like the dink shot.
• Objective kinematic analysis can enhance player performance.

Purpose of the Study:

• To introduce a MediaPipe-based deep learning tool for pickleball kinematic analysis.
• To analyze femur angle, knee angle, and wrist motion during dink shots.
• To compare kinematic differences between high-level and beginner pickleball players.

Main Methods:

• Utilized a MediaPipe-based deep learning tool for kinematic analysis.
• Recorded 14 male pickleball players performing dink shots using a GoPro camera.
• Employed T-tests and One-way ANOVA for statistical analysis of joint angles and motion.

Main Results:

• Significant differences in femur angle (p < 0.001), with advanced players showing more flexion.
• No significant difference in knee angles between player levels.
• Advanced players demonstrated continuous wrist motion post-impact (p < 0.001).

Conclusions:

• The deep learning tool provides an effective, approximate alternative to manual visual analysis for coaches.
• This technology enables rapid monitoring of kinematic parameters for targeted improvement.
• Future research should explore trunk rotation and foot positioning in various pickleball shots.