Edge AI for Real-Time HSE Compliance in Oil and Gas Operations

Abstract

This paper presents field-validated results from an edge-deployed Computer Vision Intelligence (VI) system for automated HSE compliance monitoring across multiple oilfield sites in Southeast Asia and West Africa. The solution combines advanced VI applications with state-of-the-art YOLOv8 and YOLOv11 object detection models, deployed on edge computing infrastructure. By integrating Wellsite Information Transfer Specification (WITS) protocols for drilling data, it delivers accurate, real-time surveillance and analytics for HSE compliance. The system addresses restricted/red zone management, PPE compliance, and line-of-fire detection. Results demonstrate statistically significant safety improvements validated by third-party audits, with detailed analysis supporting Goal Zero initiatives through scalable, technology-enabled safety management. The system deploys a high-performance edge device that runs a VI application. Custom models detect red zones, recognize PPE elements, and identify line-of-fire scenarios through real-time object tracking. Processing at the Edge promotes real-time violation detection of unsafe environments and HSE non-compliance at the workplace. The solution works with any hardware and cloud platform and is accessible through various network connectivity options. The application hosts proprietary AI models that are governed by physics-based rules that are set for alerting violations and provide room for customizability of site requirements. The application analyses live camera feeds and provides monitoring for safety anomalies by alerting field personnel onsite through audible and visual alarms. These instances are further available for remote analysis via a web-based application on desktop or mobile devices.