Related Experiment Video
Updated: Jan 7, 2026
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
2.0K
In situ estimation of cotton fourth internode length and height-to-node ratio using UAV-derived vegetation indices
Peter C Ngimbwa1, Denis O Kiobia1, Canicius J Mwitta2
1College of Engineering, University of Georgia, Tifton, GA, United States.
Frontiers in Plant Science
|January 1, 2026
Summary
Unmanned aerial vehicle (UAV)-derived vegetation indices (VIs) combined with machine learning (ML) algorithms accurately estimate cotton plant traits. This approach offers a more efficient alternative to traditional field measurements for precise plant growth regulator (PGR) management.
Area of Science:
- Agricultural Science
- Remote Sensing
- Machine Learning
Background:
- Accurate monitoring of cotton plant traits like height-to-node ratio and internode length is crucial for optimizing plant growth regulator (PGR) applications.
- Traditional field-based measurements are labor-intensive and may not provide timely data for effective management decisions.
Purpose of the Study:
- To investigate the efficacy of nonparametric, nonlinear machine learning (ML) algorithms using vegetation indices (VIs) from unmanned aerial vehicles (UAVs) for estimating cotton plant traits.
- To enhance the precision and efficiency of monitoring cotton growth for improved PGR application timing.
Main Methods:
- Utilized UAV-derived VIs and various ML algorithms (including Support Vector Regression and CatBoost) to estimate cotton plant height-to-node ratio and fourth internode length.
- Employed nested 5-fold cross-validation with Bayesian optimization for hyperparameter tuning and evaluated model performance using statistical tests (Friedman and Wilcoxon signed-rank).
- Applied Shapley Additive exPlanations (SHAP) to interpret the contribution of individual VIs to model predictions.
Main Results:
- Machine learning algorithms combined with UAV-derived VIs reliably estimated both height-to-node ratio (R² up to 0.8257) and fourth internode length (R² up to 0.799).
- Support Vector Regression (SVR) showed superior performance for height-to-node ratio estimation, while CatBoost excelled in predicting fourth internode length.
- The SHAP analysis provided insights into the influence of different VIs on the predictive accuracy of the models.
Conclusions:
- UAV-based VIs coupled with ML algorithms provide a consistent and accurate method for estimating key cotton plant traits.
- This technology can serve as a valuable, non-destructive alternative to traditional field measurements, enabling more efficient crop monitoring and precise PGR management.
