AI-Driven Video Analytics and Behavioral Safety Intelligence: Enhancing Human-Centered Operational Adaptation in Offshore Support Vessels

Abstract

Computer vision and artificial intelligence (AI) are increasingly used in industrial Health, Safety, and Environment (HSE) management to detect missing personal protective equipment (PPE), restricted-zone entries, and line-of-fire exposure. Early offshore and industrial deployments established the technical feasibility of real-time edge processing and automated safety monitoring (Almadhoun et al. 2020; Tamaazousti et al. 2022). A practical challenge emerges when these systems are scaled across multiple assets: large volumes of detections must be converted into operationally meaningful information. Real-time alerts can support immediate intervention, but isolated detections do not necessarily explain why a behavior recurses or whether the underlying procedure, equipment, or task design contributes to the observed deviation (Corallo and Spingardi 2024; Vukicevic et al. 2024). Traditional HSE responses may treat behavioral deviations primarily as individual compliance failures. Human Factors engineering instead recognizes that task demands, equipment interfaces, and environmental constraints shape worker behavior (Reason 1997). Safety-II similarly emphasizes the difference between prescribed work and operational practice, commonly described as Work-as-Imagined and Work-as-Done (Hollnagel 2014).