METHODS OF POSTURAL ASSESSMENT IN BADMINTON: A COMPARATIVE ANALYSIS OF CONVENTIONAL TECHNIQUES AND AI-POWERED POSE ESTIMATION

Abstract

Player posture analysis presents significant challenges in sports performance evaluation
and injury prevention, necessitating accurate and real-time detection systems. Traditional methods
of posture analysis often rely on manual observation or post-game video review, which may not
provide immediate insights or comprehensive movement data. Making use of OpenPose deep
learning framework, this study proposes a methodology for real-time player pose detection and
analysis. Initially, video footage is processed through OpenCV to extract individual frames, which
are then analyzed using a pre-trained neural network to identify key anatomical landmarks and
joint positions. Subsequently, computer vision techniques are employed to create skeletal
representations of player postures, utilizing confidence thresholds to ensure detection accuracy.
The resultant pose data is processed in real-time and can be integrated into various applications
for sports analytics, rehabilitation monitoring, and performance optimization. This model enables
coaches and medical professionals to receive immediate feedback on player movements,
facilitating proactive injury prevention and technique improvement, thereby fostering more
effective training environments for athletes at all levels.