Abstract
In order to figure out spreading of the riverine trees with remarkable growth rate at an early stage, focusing on willow, which is the object of management nationwide, this sudy validated the feasibility of automatic discrimination of the presence/absence of willow at each stage from a young tree to an adult tree. The discrimination was carried out by means of Convolutional Neural Network (CNN) using the still images captured by Unmanned Aerial Vehicle (UAV). Compared with the vegetation data acquired in advance in the field, the proportion discriminated as willow in the area where the willow is naturally present was approximately 83% in area evaluation, and 54 out of the 56 willows were counted as willow. These findings indicate that oversight of willow is less discriminated. Moreover, it was possible to discriminate young willow over 1 meter high. Furthermore, based on the discrimination results, the authors proposed a river management method to figure out the signs and the amount of proliferation of willow by coupling the work flow of UAV photogrammetry with vegetation discrimination using CNN.
