Using Unmanned Aerial Systems (UAS) and Deep Learning for Agriculture Mapping in Dubai

Abstract

As part of the 'Sustainable Future' vision, Sustainable Agriculture has become an essential pillar of the 'Food Security Strategies' formulated by the Dubai Government. Therefore, Dubai Emirate started relying on new technology to increase productivity and efficiency. In addition, agriculture applications depend on accurate land monitoring for timely food security control and support actions. However, traditional monitoring requires field works and surveys performed by experts, which is costly, slow, and sparse. Agriculture monitoring systems need sustainable land use monitoring solutions, starting with remote sensing using drone surveys for affordable, timely, and efficient agriculture mapping. Hence, Dubai Municipality is currently using Unmanned Aerial Vehicles (UAVs) for the mapping of the farming areas all over the Emirate to help to locate lands conducive to farming and creating an accurate agriculture database contributing to the decision-making process in determining the proper use of resources and areas suitable for crop growth. In this study, a novel Object detection method coupled with geospatial analysis were used in an integrated workflow to detect individual crops (accuracy = 89.7%), delineate vegetation cover (accuracy = 85.4%), and plot agriculture diseases and pests (accuracy = 83.1%) within eighteen communities in the Emirate. The detection performances of the selected deep learning method are discussed, and sample images from the datasets are used for demonstrations. The main goal is to provide specialists with an integrated solution for measuring and assessing live green vegetation cover derived from the processed images. Our results provide insight into the potential of using UAS and deep learning algorithms for sustainable agricultural mapping on a large scale.