Application of Convolutional Neural Networks (CNNs) for Work Ergonomics Analysis

Musculoskeletal disorders (MSDs) are a leading cause of work-related disability and absenteeism globally, often stemming from improper and repetitive workplace postures. Traditional ergonomic assessment methods, such as REBA and RULA, rely on manual evaluations that are inherently subjective and limited in scalability. This study presents a novel approach utilizing Convolutional Neural Networks (CNNs) to enhance the accuracy and efficiency of ergonomic risk assessments. A dataset of 1,330 workplace posture images, annotated using the REBA methodology, was analyzed through key point detection algorithms and processed with tools like Kinovea and Roboflow. The trained CNN model achieved remarkable performance metrics, including 99.9% precision, 100% recall, and 99.5% mean average precision (mAP). These results highlight the model¿s capability to classify workplace postures as correct or incorrect with high accuracy, surpassing the limitations of traditional methods. This automated approach not only eliminates subjectivity but also provides a scalable solution for MSD prevention, significantly improving workplace ergonomics. The findings of this study underscore the potential of integrating AI-driven tools with established ergonomic practices to optimize worker health and productivity in industrial environments. © 2025, Avestia Publishing. All rights reserved.

Application of Convolutional Neural Networks (CNNs) for Work Ergonomics Analysis Academic Article in Scopus