Missing plant detection and biomass estimation from 3D models generated from UAV in a vineyard

In the last decade, there has been a progressive diffusion of UAVs (Unmanned Aerial Vehicles). A computational 3D model of a vineyard can be assembled by combining low-cost UAV, RGB cameras and automated 3D reconstruction algorithms. The aim of the study was to compare two different methods of image processing in order to provide best performance for biomass estimation and missing plant detection from a 3D model of the vineyard. Both methods showed good overall accuracy, with an RMSE lower than 0.5, of biomass estimation and missing plant detection but the second method, although more precise, needed more computational time.