Propuesta de actualización de valores catastrales basado en capitalización de renta usando modelos de aprendizaje automático: Interpretación de áreas cultivadas y estimación de valor de suelo para cultivo de palma de aceite en la vereda Llano Grande San Martín - Meta

Abstract

This research proposes a system to determine cadastral values through rent capitalization using machine learning models and satellite image processing. It focuses on the identification of oil palm crops located in the Llano Grande district in San Martin, Meta (Colombia), to calculate the land value and propose a system that helps to improve the accuracy of the cadastral valuation. Neural networks and satellite images are used to determine the area of the crop and its profitability. The results show that this approach manages to reduce biases, considering the current land use, providing information that contributes to territorial planning and multipurpose cadaster management.