講演詳細

経営部門[Forest Management]

日付	2025年3月22日
開始時刻	ポスター発表
会場名	学術交流会館（ロビー）
講演番号	PD-36
発表題目	Spatial Localization of Broadleaf Species in a Mixed Forest Using UAV Multispectral Imagery and Deep Learning Spatial Localization of Broadleaf Species in a Mixed Forest Using UAV Multispectral Imagery and Deep Learning
所属	The University of Tokyo
要旨本文	Accurate spatial localization of tree species is crucial for effective forest management and ecological research. This study introduces a novel approach to segment and classify broadleaf species, including oak, in a mixed forest using UAV-acquired multispectral imagery and deep learning technique. High-resolution UAV imagery, comprising RGB and multispectral bands, was collected in eastern Hokkaido, Japan. A Mask R-CNN model was trained on annotated datasets to detect and classify individual tree crowns. The workflow integrated pre-processed UAV imagery with labeled polygons, ensuring robust model training and evaluation. Results demonstrated that incorporating multispectral bands led to higher model performance for species-level identification compared to RGB imagery alone. This study highlights the potential of combining UAV multispectral imagery with advanced deep learning methods for accurate and scalable tree species classification and forest monitoring.
著者氏名	○Nyo Me Htun1 ・ Toshiaki Owari1 ・ Satoshi N Suzuki2 ・ Kenji Fukushi1 ・ Yuuta Ishizaki1 ・ Manato Fushimi3 ・ Yamato Unno3 ・ Satoshi Kita4 ・ Ryota Konda4
著者所属	1The University of Tokyo ・ 2Hokkaido University ・ 3Tsukuba Research Institute, Sumitomo Forestry Co., Ltd. ・ 4Forest and Landscape Research Center, Sumitomo Forestry Co., Ltd.
キーワード	UAV multispectral imagery, deep learning, broadleaf species
Key word	UAV multispectral imagery, deep learning, broadleaf species