Related Experiment Video
Updated: Jul 7, 2025
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
1.4K
Utilizing Spectral, Structural and Textural Features for Estimating Oat Above-Ground Biomass Using UAV-Based
Rakshya Dhakal1, Maitiniyazi Maimaitijiang2, Jiyul Chang3
1Plant Breeding Graduate Program, University of Florida, Gainesville, FL 32608, USA.
Sensors (Basel, Switzerland)
|December 23, 2023
Summary
Unmanned Aerial Vehicles (UAVs) combined with machine learning improve oat biomass estimation by integrating spectral, structural, and textural data. This advanced phenotyping enhances crop breeding efficiency.
Area of Science:
- Agricultural Science
- Remote Sensing
- Plant Breeding
Background:
- Accurate above-ground biomass monitoring is crucial for plant breeding, but traditional methods are labor-intensive and costly.
- Unmanned Aerial Vehicles (UAVs) provide a rapid, non-destructive phenotyping solution for field plots.
- Existing Vegetation Index (VI) methods primarily use spectral data, neglecting 3D canopy structure and spatial relationships.
Purpose of the Study:
- To explore the integration of UAV multispectral imagery-derived spectral, structural, and textural features with machine learning for accurate oat biomass estimation.
- To assess the importance of canopy structural and textural features alongside spectral features.
- To compare the predictive performance of different machine learning algorithms for biomass estimation.
Main Methods:
- UAV multispectral imagery was acquired for six oat genotypes across two locations and multiple growth stages in 2020 and 2021.
- Plot-level canopy spectral, structural, and textural features were extracted from the imagery.
- Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), and Random Forest Regression (RFR) were employed to estimate biomass.
Main Results:
- Canopy structural and textural features were identified as important indicators for oat biomass estimation, complementing spectral data.
- Combining spectral, structural, and textural features significantly enhanced biomass estimation accuracy compared to using single feature types.
- Machine learning algorithms demonstrated strong predictive ability, with Random Forest Regression (RFR) achieving the highest accuracy (R² = 0.926, RMSE% = 15.97%).
Conclusions:
- UAV-based multi-feature fusion with machine learning offers a promising approach for accurate above-ground biomass estimation in oat breeding nurseries.
- This integrated method can significantly improve the efficiency of oat breeding programs through advanced phenotyping.
- The findings support the adoption of UAV-based phenotyping for enhanced crop management practices.
