A Machine Learning Framework for Real-Time Personalized Ergonomic Pose Analysis

This paper presents a novel methodology for real-time ergonomic pose analysis using volumetric video data, specifically leveraging 3D point clouds to enhance pose inference from multiple angles. The system employs a personalized deep learning classifier trained on user-labeled poses, demonstrating effective real-time skeletal labeling during load-lifting tasks captured by RGB-D cameras. This approach offers a scalable solution for ergonomic evaluations, addressing critical limitations in traditional 2D pose estimation and enhancing workplace safety monitoring.