Mapping Tree Species Using Deep Learning and Airborne Observation Data

Abstract

In this study, we introduce a CNN (convolutional neural network) which mimics professional interpreters’ manual techniques. Using simultaneously acquired airborne imageries and LiDAR data, we attempt to reproduce the 3D knowledge of tree shape, which interpreters potentially make use of. Geospatial features which support interpretation are also used as inputs to the CNN. Inspired by the interpreters’ techniques, we propose a unified approach that integrates these datasets in a shallow layer. With the proposed CNN, we show that the CNN works robustly.