Drones in high resolution land use assessment using artificial intelligence

Abstract

Unmanned aerial vehicle (UAV) remote sensing has enormous potential for land use land cover mapping in complicated landscapes and locations. This is given to the ultra-high-resolution images that can be collected at low altitudes using these vehicles. Due to cargo capacity limits, off-the-shelf digital cameras are widely employed on medium- and small-sized UAVs. When it comes to land use and land cover mapping, the low spectral resolution that digital cameras have can be a disadvantage, but it can be solved with the help of texture features and sophisticated machine learning classifiers. Although numerous machine learning techniques are routinely employed in satellite remote sensing applications, there is little information available on how it is utilized for UAV image classification, and it has not yet been investigated whether or not these algorithms are beneficial for classifying images with low spectral resolution. The applicability of a variety of machine learning algorithms, such as Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN), was evaluated in order to determine how accurately they could differentiate between the various types of land use and land covers found in the agricultural farms located at Banaras Hindu University, Varanasi. © 2025 Elsevier Ltd. All rights reserved.