RGB-based 3D reconstruction for organ-level maize phenotyping under contrasting phosphorus treatments

RGB-based 3D reconstruction for organ-level maize phenotyping under contrasting phosphorus treatments OVERVIEW This repository contains the raw reconstructed 3D point-cloud data used in the study “RGB-based 3D reconstruction for organ-level maize phenotyping under contrasting phosphorus treatments.” The dataset consists of RGB-derived dense 3D point clouds of maize plants grown under two phosphorus (P) treatments: High-P: plants grown under high-phosphorus conditions. Low-P: plants grown under low-phosphorus conditions. The repository also contains four artificial maize plant samples used to validate the point-cloud-based phenotyping workflow. The data are provided to support transparency, reproducibility, independent re-analysis, and further methodological development in 3D plant phenotyping. REPOSITORY STRUCTURE The dataset is organized into three main folders: HighP plants LowP plants artificial plants The HighP plants folder contains raw reconstructed 3D point clouds of individual maize plants grown under the High-P treatment. The LowP plants folder contains corresponding data for plants grown under the Low-P treatment. The artificial plants folder contains the 3D point-cloud data of four artificial maize plant samples used to validate the phenotyping workflow. HIGH-P DATASET The HighP plants folder contains raw reconstructed 3D point-cloud data for individual maize plants grown under the High-P treatment. Each subfolder corresponds to one experimental plant. The High-P dataset contains the following plant folders: plot_1_h plot_10_h plot_18_h plot_36_h plot_45_h plot_53_h plot_69_h plot_74_h plot_82_h Folder names ending in _h indicate samples belonging to the High-P treatment. Each plant-specific folder contains the corresponding raw reconstructed 3D point-cloud file in .ply format. For example, plot_1_h contains pot1_highp.ply. The .ply file represents the three-dimensional geometry of the corresponding maize plant reconstructed from the RGB image sequence. LOW-P DATASET The LowP plants folder contains raw reconstructed 3D point-cloud data for individual maize plants grown under the Low-P treatment. Each subfolder corresponds to one experimental plant. The Low-P dataset includes: plot_4_l plot_5_l plot_22_l plot_39_l plot_49_l plot_72_l plot_75_l plot_89_l plot_108_l Folder names ending in _l indicate samples belonging to the Low-P treatment. Each plant-specific folder contains the corresponding raw reconstructed 3D point-cloud file in .ply format. ARTIFICIAL PLANT DATASET The artificial plants folder contains 3D point-cloud data for four artificial maize plant samples. These samples were used to validate the point-cloud-based phenotyping workflow. Because the artificial plants provide stable reference geometries, they can be used to evaluate the accuracy of morphological measurements extracted from the reconstructed 3D point clouds. FILE FORMAT The main file format used in this repository is .ply. PLY is a commonly used format for storing three-dimensional point-cloud and surface data. Each .ply file contains the reconstructed three-dimensional geometry of an individual plant sample. The point-cloud files can be viewed and processed using software such as CloudCompare, MeshLab, Open3D, ParaView, or other software capable of reading .ply files. NAMING CONVENTION Folder names indicate the phosphorus treatment associated with each plant sample: Folders ending in _h correspond to the High-P treatment. Folders ending in _l correspond to the Low-P treatment. For example, plot_1_h represents a sample from the High-P treatment, whereas plot_4_l represents a sample from the Low-P treatment. The numerical part of each folder name identifies the corresponding experimental plant or plot. WHAT EACH PLANT FOLDER CONTAINS Each plant-specific folder contains the raw reconstructed point-cloud file for one individual maize plant. For example, plot_1_h contains pot1_highp.ply. The .ply file represents the three-dimensional plant geometry reconstructed from the corresponding RGB image sequence. Where available, a reference or preview image may also be included to assist with visual inspection of the corresponding sample. Depending on the repository or file-browser interface, supported image or 3D files may be displayed directly as previews. IMPORTANT: THESE ARE RAW POINT-CLOUD DATA The point-cloud files provided in this repository represent raw reconstructed 3D data and should not be interpreted as final, analysis-ready phenotypic measurements. To reproduce the measurements and results reported in the associated manuscript, the raw point clouds should be processed using the same workflow and methodological settings applied in the study. REQUIRED PROCESSING STEPS 1. Scaling and calibration The reconstructed point clouds must first be converted to the appropriate physical scale before quantitative morphological measurements are calculated. 2. Point-cloud cleaning and preparation Reconstruction noise, background points, and other non-plant points should be removed where necessary before further analysis. 3. Organ-level segmentation The aboveground plant point cloud must be segmented into the relevant plant organs, including individual leaves and stem regions. 4. Organ identification Individual leaves should be identified and numbered consistently according to the leaf-numbering convention used in the associated study. 5. Trait extraction After segmentation, the processed point clouds can be used to calculate structural traits such as leaf length, maximum leaf width, mean leaf width, leaf area, mean midrib angle, and stem width, depending on the analysis. 6. Statistical analysis The extracted structural traits can subsequently be compared between the High-P and Low-P treatments using the statistical procedures described in the associated manuscript. SUGGESTED PROCESSING WORKFLOW RGB image sequence↓3D reconstruction↓Raw dense point cloud in .ply format↓Scaling and calibration↓Point-cloud cleaning↓Organ-level segmentation↓Leaf and stem identification↓Trait extraction↓Statistical analysis↓Results reported in the manuscript ASSOCIATED STUDY Title: RGB-based 3D reconstruction for organ-level maize phenotyping under contrasting phosphorus treatments CONTACT INFORMATION Khandoker AhammadDoctoral ResearcherInstitute of Agricultural EngineeringAgricultural Engineering in the Tropics and SubtropicsUniversity of Hohenheim Garbenstrasse 970599 Stuttgart, Germany Telephone: [phone]Email: [e-mail]