Advancing workplace safety through intelligent monitoring: Applications of artificial intelligence and computer vision in high-risk industrial environments

High-risk industrial environments have continued to experience elevated workplace injuries, in part because traditional manual safety assessments still face major drawbacks related to subjectivity, limited scalability, and difficulty adapting to complex and changing environmental contexts. The primary objective of this research is to systematically investigate and address critical gaps in feasibility, scalability, contextual enhancement, and adaptive intelligence for computer vision-based safety and ergonomic risk assessment across high-risk industrial environments. The research methodology follows a systematic four-phase process, each contributing critically to the overall objective. Phase one establishes computer vision feasibility for personal protective equipment detection in steel manufacturing environments using labeled image datasets and cross-validation techniques. Phase two develops an integrated unmanned aerial vehicle (UAV)-based computer vision framework using pose estimation algorithms for automated ergonomic risk assessment, validated through construction site deployment. Phase three creates the Elevated Construction Ergonomic Risk Index (ECERI) using multi-tier validation methodology, including theoretical proofs, computational simulations, and empirical expert judgment comparisons to enhance traditional Rapid Entire Body Assessment (REBA) with environmental context factors. Phase four implements the Self-Organizing Fuzzy Inference System (SOFIS) incorporating dual uncertainty quantification through information-theoretic frameworks and Monte Carlo methods with expert-in-the-loop validation. The research demonstrates successful computer vision implementation in challenging industrial environments, validates scalable UAV frameworks for comprehensive ergonomic monitoring, establishes context-aware assessment methods improving traditional approaches, and creates adaptive intelligent systems with uncertainty awareness and continuous learning capabilities. These findings provide validated solutions addressing each identified gap in current safety monitoring approaches, contributing practical tools and theoretical advancement for improved occupational safety monitoring systems across high-risk industries.