3D morphological benchmark datasets for accelerating plant phenotyping studies

2020 Fiscal Year Final Research Report

• Project/Area Number: 19K15946
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 41040:Agricultural environmental engineering and agricultural information engineering-related
• Research Institution: Kyushu University
• Principal Investigator: Noshita Koji 九州大学, 理学研究院, 助教 (10758494)
• Project Period (FY): 2019-04-01 – 2021-03-31
• Keywords: 植物フェノタイピング / 数理モデル / 形態測定学 / 画像解析

Outline of Final Research Achievements

I worked on the creation of a 3D morphological dataset of plant structure. I acquired multi-view images by using a photogrammetry studio with multiple digital cameras, created annotation data for the obtained multi-view images, reconstructed the 3D images using the multi-view images, estimated phenotypic values by analyzing the reconstructed point cloud data, and developed a method for properties that have been difficult to quantify. As a result, we created a 3D plant morphological dataset, especially for traits related mainly to leaves of soybean. In addition, we were able to demonstrate the value of the dataset in the development of a novel mathematical model-based measurement method and a model.

Free Research Field

数理生物学，形態測定学，農業情報学

Academic Significance and Societal Importance of the Research Achievements

植物地上部の３次元形態データセットの作成に取り組んだ．これにより，新規に開発する計測方法の評価や機械学習モデルの訓練データとしての活用が見込める．また，従来計測が困難であった性質の評価手法の開発にもつながった．すなわち，詳細な計測データ，アノテーションデータ，形質データを組み合わせたデータセットを作成することで，新規に提案される計測手法の有用性を検証できる基盤の構築や新しい手法の開発の支援をおこなうことが可能になると考えられる．