AI-powered gas leak detection system

This article examines the problem of gas leaks in a gas field and proposes a solution in the form of a gas leak detection system. One component of this system is computer vision-based gas leak detection using infrared cameras. Subject of research: automation of the process of detecting gas leaks in a gas field using computer vision algorithms. Purpose of research: developing and conducting experimental studies to determine the effectiveness of a computer vision-based gas leak detection system. Research methods: computer vision and deep learning methods applied to the analysis of data obtained during experimental studies at a laboratory test bench. Object of research: a gas leak process simulated in a laboratory setting and an automatic leak detection subsystem based on computer vision algorithms. Research findings: a generated and labeled dataset of gas leak recordings; a trained computer vision model based on this dataset; experimental testing of a subsystem capable of detecting gas leaks. The results confirm the fundamental viability of the proposed approach within the first block of the integrated system.