A Low-Cost Photogrammetry System for 3D Plant Modeling and Phenotyping

TL;DR

The paper tackles the barrier of expensive, complex phenotyping by introducing an open-source, low-cost photogrammetry system based on structure-from-motion (SfM) and a turntable to generate rich 3D plant models. It combines off-the-shelf hardware (Raspberry Pi, four 64MP cameras, motorized turntable) with an automated software pipeline (image acquisition, COLMAP-based reconstruction, and Open3D-based processing) to extract quantitative plant traits from 3D point clouds. Key contributions include a complete hardware description, open software for push-button data generation, a semi-automated segmentation pipeline, and a suite of phenotypic metrics (height, radius, convex hull, ground-cover area, and leaf angles) that correlate with expert canopy classifications in wheat. The system demonstrates cost-effective, scalable 3D phenotyping with potential for breeding programs and field adaptation, with public data and code available for reproducibility and further development.

Abstract

We present an open-source, low-cost photogrammetry system for 3D plant modeling and phenotyping. The system uses a structure-from-motion approach to reconstruct 3D representations of the plants via point clouds. Using wheat as an example, we demonstrate how various phenotypic traits can be computed easily from the point clouds. These include standard measurements such as plant height and radius, as well as features that would be more cumbersome to measure by hand, such as leaf angles and convex hull. We further demonstrate the utility of the system through the investigation of specific metrics that may yield objective classifications of erectophile versus planophile wheat canopy architectures.

Figures (16)

• Figure 1: (Left) A view of turntable with an example wheat plant and blue backdrop. In the foreground three of the four auto-focus cameras can be seen. (Right) The same system viewed from the side of the backdrop, showing the aluminum extrusion stand with the four cameras, the Raspberry Pi, and lights for additional illumination.
• Figure 2: Schematic of the modified wiring required to drive the the turntable stepper motor via the RPi and Adafruit DC & Stepper Motor HAT.
• Figure 3: Screenshot of the system's graphical user interface, showing the settings the user can adjust, including which cameras to use, how far the turntable rotates between images taken and the wait time after rotation for any movements of the plant to settle.
• Figure 4: Flowchart of individual steps involved in collecting images with the photogrammetry system.
• Figure 5: Images captured from four different perspectives by cameras A through D (top row) and the corresponding binary images (bottom row). Camera A represents the highest vantage point, with each subsequent camera (B to D) capturing from progressively lower perspectives. The binary images (bottom row) show regions where no features will be extracted in black.