OGInfra: Geolocating Oil & Gas Infrastructure using Remote Sensing based Active Fire Data

TL;DR

This paper introduces a novel deep learning-based method for accurately geolocating oil and gas infrastructure using active fire data from NASA, achieving over 90% accuracy.

Contribution

It presents a new automated approach combining remote sensing data and deep learning for precise infrastructure geolocation.

Findings

Achieved 90.68% top accuracy with ResNet101

Demonstrated effectiveness of active fire data for infrastructure localization

Proposed a scalable method for oil & gas infrastructure mapping

Abstract

Remote sensing has become a crucial part of our daily lives, whether it be from triangulating our location using GPS or providing us with a weather forecast. It has multiple applications in domains such as military, socio-economical, commercial, and even in supporting humanitarian efforts. This work proposes a novel technique for the automated geo-location of Oil & Gas infrastructure with the use of Active Fire Data from the NASA FIRMS data repository & Deep Learning techniques; achieving a top accuracy of 90.68% with the use of ResNet101.