Comparative Analysis of Rgb and Depth Data For Enhanced Functional Fitness Assessments and Accurate Human Pose Estimation

This study presents an automated solution for human pose estimation and skeletal tracking in functional fitness assessments for older adults. Focusing on the Chair Sit-and-Reach exercise from the Fuller- ton Battery, it automates distance and angle calculations between body segments. Two Kinect-based implementations were developed: (i) Medi- aPipe’s RGB-based 2D pose detection and (ii) PyKinect’s 3D skeletal tracking with depth data. User tests showed that MediaPipe exhibited higher accuracy in distance measurements, reinforcing its potential to replace manual assessments and reduce human error.