Related Experiment Video
Updated: Aug 6, 2026
06:28
High Throughput Image-Based Phenotyping for Determining Morphological and Physiological Responses to Single and Combined Stresses in Potato
Published on: June 7, 2024
F2DMAS: a smartphone video-based 3D phenotyping workflow for potted plants in complex backgrounds
Jian Fang1,2,3,4, Nengfu Xie1,3,4, Yane Duan2
1Agricultural Information Institute, Chinese Academy of Agricultural Sciences, Beijing, China.
Frontiers in Plant Science
|July 23, 2026
Summary
This study introduces F2DMAS, an automated 3D plant phenotyping workflow using smartphone videos. It enables accurate plant trait extraction from complex environments, offering a practical solution for low-cost phenotyping.
Area of Science:
- Agricultural Science
- Computer Vision
- Plant Biology
Background:
- Accurate 3D plant structural information is crucial for phenotyping.
- Conventional 3D reconstruction methods struggle with complex backgrounds and motion blur in practical settings like greenhouses.
- This limits low-cost and automated plant phenotyping.
Purpose of the Study:
- To develop an automated 3D plant phenotyping workflow using consumer-grade smartphone videos.
- To overcome limitations of conventional methods in non-ideal acquisition conditions.
- To enable accurate extraction of plant phenotypic traits.
Main Methods:
- F2DMAS workflow converts smartphone videos to image sequences, filtering motion blur.
- A frequency-spatial plant segmentation module (FSAM3) separates plants from complex backgrounds without annotated data.
- Reconstruction uses 2D Gaussian Splatting, followed by TSDF-based meshing and scale recovery.
Main Results:
- F2DMAS achieved stable 3D plant reconstruction under non-ideal conditions (PSNR: 31.09, SSIM: 0.9711).
- The workflow significantly reduced mesh extraction time and improved reconstruction quality compared to baseline methods.
- Extracted phenotypic traits showed high agreement with manual measurements (R²: 0.90–0.99).
Conclusions:
- F2DMAS provides an end-to-end solution for 3D plant phenotyping from smartphone video acquisition to trait extraction.
- The method is practical and deployable for greenhouse, seedling, and potted plant experiments.
- It offers a robust and cost-effective approach for automated plant phenotyping.
