To estimate trunk volume and be visualization inside forest, supplemental software tools were developed for segmentation of tree components and measurement of size parameters by using of the latest machine learning for 3D point cloud. Point cloud of two plantation forests obtained in Fukushima Prefecture was preprocessed by the developed remote sensing tool of forest (las2tree) and manual segmentation separating into tree trunk and other components.Customized PointNet++ performed automatic segmentation for tree components. Accuracy of the segmentation was favorable for the future forest modeling process.
