講演詳細

経営部門[Forest Management]

日付	ポスター発表
会場名	（学生ポスター賞の審査対象）
講演番号	P-085
発表題目	Discriminating conifer and broadleaf cover in an uneven-aged forest using UAV imagery and machine learning Discriminating conifer and broadleaf cover in an uneven-aged forest using UAV imagery and machine learning
要旨本文	This study aimed to explore the feasibility of UAV imagery to discriminate coniferous and broadleaf cover in an uneven-aged mixed forest by applying machine learning classification algorithms. Our study area was Sub-compartment 42B in the University of Tokyo Hokkaido Forest (90.3 ha). The aerial images were acquired using a DJI-Inspire 2 UAV platform in August 2022. We analyzed the RGB information of UAV images over the study area, through semantic segmentation schemes using Random Forest and U-Net models. 80% of the dataset were used for training while 20% for validation for both models. Our results showed that the validation accuracy of U-Net was over 90%, while Random Forest failed to distinguish conifer canopies. Our case study revealed the integration of UAV imagery and U-Net model was more reliable to segment the conifer and broadleaf cover. Moreover, the findings highlight an applicable methodology for describing the dominated tree species groups in uneven-aged mixed forests.
著者氏名	○Nyo Me, Htun1 ・ Owari, Toshiaki2 ・ Tsuyuki, Satoshi1 ・ Hiroshima, Takuya1
著者所属	1The University of Tokyo ・ 2The University of Tokyo
キーワード	conifer, broadleaf, UAV imagery, uneven-aged forest, machine learning
Key word	conifer, broadleaf, UAV imagery, uneven-aged forest, machine learning