Automatic detection and quantification of wild game crop damage using an unmanned aerial vehicle (UAV) equipped with an optical sensor payload: A case study in wheat

Abstract

Wildlife-induced damage of agricultural crops is an unfavorable consequence of elevated population densities of wild animals, especially wild boars. For the purposes of financial compensations for crop damage, provided by either governments or hunters responsible for game numbers, it is necessary to precisely assess the range of damage and temporal change. The use of an unmanned aerial vehicle (UAV) with an optical sensor payload represents a potential method of obtaining data of crop conditions without the necessity to enter the field and increase the damage. We propose a novel method for delineation of damaged areas via automatic segmentation of the crop field. Our method is based on photogrammetric reconstruction of the various crop heights within the field through the use of Structure from Motion technique with subsequent automatic classification. In this case study of wheat, the range of damage was estimated with an accuracy of 99.5% and 99.3% using field global navigation satellite system (GNSS) measurements and classification of an orthomosaic generated from UAV-based imagery, respectively.