Abstract: Marine oil and gas infrastructures are continuously exposed to the risk of hydrocarbon spills, necessitating rapid, automated surveillance frameworks to mitigate ecological catastrophes. This paper presents a multi-modal, automated real-time environmental monitoring system that fuses computer vision with localized Internet of a Things (IoT) biochemical telemetry. The detection architecture is powered by an optimized lightweight deep learning network (YOLO) trained on a heterogeneous dataset compiling high-resolution Unmanned Aerial Vehicle (UAV) RGB imagery and Synthetic Aperture Radar (SAR) Ground Range Detected (GRD) data from NASA Earthdata and the Copernicus Sentinel Data Hub. Simultaneously, an internet telemetry gateway continuously processes aquatic health parameters, including Water pH, Turbidity, and Dissolved Oxygen (DO). The backend infrastructure is built using the Django REST Framework to handle asynchronous image preprocessing, model inference, and a localized alert generation mechanism. Real-time visualization is rendered via a decoupled React.js single-page web dashboard. Empirically, the AI classification layer achieved an overall validation accuracy of 94.6% and an F1-score of 0.945, demonstrating a frame processing latency of 22 milliseconds. Combined with a systemic end-to-end telemetry response time of 1.84 seconds, the proposed framework offers an efficient, early-warning operational pipeline for environmental regulatory authorities and asset operators.
Keywords: Oil Spill Detection, You Only Look Once (YOLO), Internet of a Things (IoT), Multi-Modal Fusion, Remote Sensing, Environmental Monitoring Dashboard.
Title: Real-Time Environmental Monitoring Dashboard and Multi-Modal AI Detection System for Marine Oil and Gas Facilities
Author: Nkue Dumka, Emmanuel Iwok Essien, Uliang Ibia Ibia, AniebietAbasi Prosper Daniel, Hameedat Bashir, Iwundu Elizabeth Chigozirim
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Vol. 14, Issue 3, July 2026 - September 2026
Page No: 17-23
Research Publish Journals
Website: www.researchpublish.com
Published Date: 11-July-2026
