Benefit and cost analysis for utilizing machine learning in geospatial projects

Abstract

This paper aims to study the feasibility of applying deep learning techniques in the geospatial application domain. Two geospatial projects were economically analysed. The first project was to produce land use maps from satellite imagery for the current and past years using three machine learning algorithms. Based on the produced historical maps, scenario predictions of the land use maps were generated [1]. The second geospatial project used a deep learning object detection technique to count the total numbers of palm trees automatically and develop a geospatial database that has the exact coordinate location per each palm tree distributed in the Kingdom of Bahrain [2]. The technical, administrative, and financial costs were identified as well as the benefit in the monetary value per each project. As a result, benefit-cost analysis was conducted to evaluate both projects economically.