A Study on tree species discrimination using machine learning in forestry

Abstract

As the present state of forestry in Japan, the period of full-scale utilization of planted forests is approaching; consolidation of forest practices aimed at the stable and efficient supply of domestic timber is being promoted. Visualization of forest boundaries is important, however it takes a lot of cost and time because forest surveys are conducted by human’s exploration. This study proposes an efficient tree species identification method using deep learning method from aerial images of forests using drones for the purpose of visualizing forest boundaries. The effectiveness of the proposed method is confirmed by experiments using actual images taken by drones.