Collaboration

RIoT Research Center, IUB & Department of Crop Botany, Bangladesh Agricultural University

Traditionally, manual tree species identification is performed by travelling through the trails, having experts identify the trees, and documenting their identification. This indicates the necessity for an automatic tree species identification system for botanical survey because the traditional method is tedious, time-consuming, and dependent on the availability of an expert. Several methods proposed for this purpose, either using remote sensing data or digital images captured using a digital camera or UAV. The restrictions of a manual technique are still present in digital camera-based systems, which require an operator to carry the camera and capture images. UAV-based technology has recently been found useful in forestry applications, particularly due to its low cost and high resolution (a few centimeters per pixel). In this study, we propose a lost cost, easy-to-use and time-efficient tree species identification system utilizing deep learning and a UAV-captured RGB-image for botanical surveys.