AI-Based Pose Detection for Ergonomic Risk Screening in Manual Lifting Tasks

Abstract

This study analyzes lifting posture using Artificial Intelligence–based pose detection technology through the APECS: Body Posture Evaluation application as part of the Industry 5.0 framework, which emphasizes human-technology collaboration. A descriptive qualitative method was applied to five Respondents, each photographed once during the initial phase of lifting a load to assess body alignment and joint angles. Results show that four respondents demonstrated ergonomic posture, with an upright back position and proportional knee bending, while one respondents exhibited a non-ergonomic posture with a 37° spinal alignment angle that potentially increases musculoskeletal injury risk. APECS proved useful for providing rapid and objective visualization of posture quality, although it is limited to single-frame analysis and cannot capture dynamic movement changes which makes it most suitable for routine spot-check audits, pre-task coaching, and supporting decisions such as identifying high-risk individuals, prioritizing refresher training, and standardizing simple ergonomic checkpoints rather than diagnosing full movement patterns. Overall, AI-based pose detection shows potential as an effective tool for monitoring workplace posture and improving safety in alignment with ergonomic principles and Industry 5.0 developments by enabling quicker supervisory feedback loops and more consistent documentation of posture quality in regular safety programs.