Artificial Intelligence for Gas Leak Detection with Thermal Cameras and Metal Oxide Semiconductor Sensors

• Edmilson Sanches CPQD http://orcid.org/0009-0009-2235-3417
• Fábio Augusto CPQD https://orcid.org/0009-0000-7288-3525
• Pauline Silveira CPQD
• Bernardo Feijó Junqueira CPQD
• Dimas A. M. Lemes PUC-Campinas
• José Picolo PUC-Campinas
• Guilherme Ribeiro Sales CPQD
• Valentino Corso CPQD
• Cides S. Bezerra CPQD https://orcid.org/0000-0002-3602-1909

Abstract

Early detection of gas leaks is crucial for safety and efficiency in oil platforms and refineries. The presence of various hazardous gases, often imperceptible to human senses, poses significant risks. AI-powered solutions can effectively monitor for gas leaks, improving safety and ensuring efficient operations. In this work we proposed a modular architecture effectively combines tabular data from gas sensors and spatial information from thermal images using a variety of backbones, including MobileNet. By employing dense layers and an optimized training strategy, we achieved state-of-the-art performance, with 100% accuracy, demonstrating the effectiveness of our approach for gas leakage detection.

Keywords: Gas Leak Detection, AI-powered Solution, Multimodal Data Fusion