A spatial data image processing model using unmanned aerial vehicles and RGB imagery for crop farming on small-scale subsistence farms

This paper presents a spatial data image processing model using unmanned aerial vehicles (UAVs) equipped with RGB cameras for enhancing agricultural productivity on small-scale subsistence farms in semi-arid regions of Botswana. The model leverages high-resolution UAV imagery (0.19 cm/ pixel) to transform raw, non-spatial complex data into actionable geospatial information. It integrates machine learning algorithms, including unsupervised ISODATA and supervised support vector machine (SVM) classifiers, with proposed new RGB-based vegetation indices, to extract detailed information on crop types, land use, and coverage. The research addresses the challenges faced by smallholder farmers, such as limited access to advanced technologies and fluctuating environmental conditions. This model facilitates improved agricultural decisions and productivity by providing a cost-effective and accurate method for monitoring and managing crops through localised data capture and advanced data processing techniques. Results demonstrate an overall classification accuracy of 82.5% with the ISODATA algorithm, proving its utility in precision agriculture tailored for resource-limited settings. The study underscores the potential of integrating UAV technology with machine learning to support sustainable agriculture in developing regions.