Digital Phenotyping in Plant Breeding: Evaluating Relative Maturity, Stand Count, and Plant Height in Dry Beans via RGB Drone-Based Imagery and Deep Learning Approaches

Abstract

The datasets included were collected from the MSU dry bean breeding project sites in planting seasons 2020, 2021 and 2022. Specific planting season includes the digital surface models (DSM), digital terrain model (DTM) or point cloud (PC) from above and soil level generated using the raw images, plot boundary delimitations (.shp), and the ground-truth notes to plant height (PH) estimation. /2020: This directory contains 4 folders /a._Ground_notes 1. 2020 N&B Raw.xlsx Ground truth notes collected in 2020 at SVREC location for Black and Navy bean market classes. /b._Shapefiles Plot boundaries files (.shp) and field area from 2020 SVREC location containing plot level information using the breeding program metadata. /c._DSM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Surface Model (TIFF images) collected from vegetation containing readable EXIF headers with image metadata. /d._PC 1. Image naming structure: `Date-of-flight _ _ /` Point Cloud (LAZ files) collected from the vegetation containing data points from above ground used to perform the plot reconstruction analysis. ################################################################################### /2021_HURON: This directory contains 6 folders /a._Ground_notes 1. 2021 N&B Raw.xlsx Ground truth notes collected in 2021 at HURON location for Black and Navy bean market classes. /b._Shapefiles Plot boundaries files (.shp) and field area from 2021 HURON location containing plot level information using the breeding program metadata. /c._DSM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Surface Model (TIFF images) collected from vegetation containing readable EXIF headers with image metadata. /d._DTM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Terrain Model (TIFF image) collected before vegetation established containing readable EXIF headers with image metadata. /e._PC_veg 1. Image naming structure: `Date-of-flight _ _ _ /` Point Cloud (LAZ files) collected from the vegetation containing data points from above ground used to perform the plot reconstruction analysis. /f._PC_soil 1. Image naming structure: `Date-of-flight _ _ _ /` Point Cloud (LAZ files) collected from the bare soil containing data points from ground level used to perform the plot reconstruction analysis. ################################################################################### /2021_SVREC: This directory contains 6 folders /a._Ground_notes 1. 2021 N&B Raw.xlsx Ground truth notes collected in 2021 at SVREC location for Black and Navy bean market classes. /b._Shapefiles Plot boundaries files (.shp) and field area from 2021 SVREC location containing plot level information using the breeding program metadata. /c._DSM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Surface Model (TIFF images) collected from vegetation containing readable EXIF headers with image metadata. /d._DTM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Terrain Model (TIFF image) collected before vegetation established containing readable EXIF headers with image metadata. /e._PC_veg 1. Image naming structure: `Date-of-flight _ _ _ /` Point Cloud (LAZ files) collected from the vegetation containing data points from above ground used to perform the plot reconstruction analysis. /f._PC_soil 1. Image naming structure: `Date-of-flight _ _ _ /` Point Cloud (LAZ files) collected from the bare soil containing data points from ground level used to perform the plot reconstruction analysis. ################################################################################### /2022/SVREC: This directory contains 4 folders /a._Ground_notes 1. 2022 N&B Raw.xlsx Ground truth notes collected in 2022 at SVREC location for Black and Navy bean market classes. /b._Shapefiles Plot boundaries files (.shp) and field area from 2022 SVREC location containing plot level information using the breeding program metadata. /c._DSM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Surface Model (TIFF images) collected from vegetation containing readable EXIF headers with image metadata. /d._PC 1. Image naming structure: `Date-of-flight _ _ /` Point Cloud (LAZ files) collected from the vegetation containing data points from above ground used to perform the plot reconstruction analysis. ################################################################################### /2022/HURON: This directory contains 4 folders /a._Ground_notes 1. 2022 N&B Raw.xlsx Ground truth notes collected in 2022 at HURON location for Black and Navy bean market classes. /b._Shapefiles Plot boundaries files (.shp) and field area from 2022 HURON location containing plot level information using the breeding program metadata. /c._DSM 1. Image naming structure: `Date-of-flight _ _ _ /` Digital Surface Model (TIFF images) collected from vegetation containing readable EXIF headers with image metadata. /d._PC 1. Image naming structure: `Date-of-flight _ _ /` Point Cloud (LAZ files) collected from the vegetation containing data points from above ground used to perform the plot reconstruction analysis. For general information or questions about the UAS-based imagery analysis to estimate plant height (PH), please contact leo.agroufv@gmail.com (Leonardo Volpato). A pipeline software tool and scripts for data extraction and analysis were developed to accomplish the activities ranging from image capture to statistical analysis of extracted features. The plant height (PlantHeightR) R shiny software can be accessed at https://github.com/msudrybeanbreeding/PlantHeightR The UAS-based PH date scripts and processes used to perform the image analyses and trait extract are available at https://github.com/msudrybeanbreeding?tab=repositories.

Plant phenotyping relevance

RGBドローン画像・DSM・点群から植物高などを推定するデータセットと解析パイプライン、R Shinyソフトウェアを提供しており、植物表現型の取得・抽出手法が中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.