Vision-Based Real-Time Ergonomic Detection in SMEs

Abstract

As highlighted in this paper, the project development was a real-time ergonomic risk detection system with an eye on small and medium-sized enterprises (SMEs). The ultimate purpose of the study is to develop an inexpensive, visual sensor-based surveillance system that could improve the working environment safety through identification of unfavorable postures that can cause working-related musculoskeletal disorder (WMSD). The system integrates webcam data with Python-anchored pose-estimation software to locate body points and compute body joint angles. Deviations in posture get measured and in case the angle of extension surpasses 20 o angle with the neutral, warnings can be seen and heard in real time to trigger corrections. Attention to ergonomic conflict score is further done through Rapid upper limb assessment (RULA) basis of scoring and results saved in CSV format which thereafter analyzed. The prototype system was logically developed, set up and tested under an SME-like construct. The experimental data showed that the system can recognize unsafe postures correctly and the feedback was timely as well as effective at encouraging employees to correct themselves in behavioral patterns. The data recorded also indicated the trend of ergonomic risk during working time periods and this proved the effectiveness of the system. The work is planned to be enhanced in the future to fine-tune operation under lighting conditions changes, be capable of detection multiple workers simultaneously, and test the algorithm on more diverse SME configurations. This paper shows the possibility of inexpensive, real-time, vision-based solutions in enhancing ergonomic operations and boosting the safety and security of employees in SMEs.