Session 2: AI-driven wheat yield and protein content forecasting using UAV remote sensing

Presentation Type

Oral

Student

Yes

Track

Other

Abstract

Preharvest forecasting of wheat grain yield, test weight and protein content is critical in terms of in season decision making and field management practices, as well as field-based high-throughput phenotyping toward enhanced yield and grain quality. In recent years, the rapid advancement of Unmanned Aerial Vehicle (UAV) and sensor technologies enabled high-resolution spatial, spectral, and temporal data collection with a lower cost. Coupled with cutting-edge artificial intelligence and deep learning (AI/DL) algorithms, UAV remote sensing has become an important tool in a variety of agricultural applications. This study aims to investigate the potential of UAV-based multitemporal multispectral data for preharvest wheat yield, test weight and protein content estimation under the framework of AI/DL. UAV-based multispectral images were collected throughout the 2022 winter wheat growing season over seven experimental winter wheat fields across South Dakota, USA. Plot-level canopy spectral and texture features were derived from UAV multispectral imagery. Traditional machine learning approaches such as Partial Least Squares Regression, Support Vector Regression, and Random Forest Regression were employed to develop prediction models using plot-level averaged spectral and texture features. Additionally, deep learning methods such as Convolutional Neural Networks (CNN) and hybrid CNN and Long-Short Term Memory (CNN-LSTM) were also implemented using plot-level reflectance imagery as input for prediction model development. This research highlights the potential of coupling high-resolution UAV remote sensing with cutting-edge AI/DL in predicting wheat yield, test wight and protein content. The results from this work deliver valuable insights for high-throughput phenotyping and crop field management with high spatial precision.

Key words: artificial intelligence (AI), deep learning (DL), unmanned aerial vehicle (UAV), remote sensing, winter wheat, yield prediction

Start Date

2-7-2023 9:50 AM

End Date

2-7-2023 10:50 AM

This document is currently not available here.

Share

COinS
 
Feb 7th, 9:50 AM Feb 7th, 10:50 AM

Session 2: AI-driven wheat yield and protein content forecasting using UAV remote sensing

Pheasant Room 253 A/B

Preharvest forecasting of wheat grain yield, test weight and protein content is critical in terms of in season decision making and field management practices, as well as field-based high-throughput phenotyping toward enhanced yield and grain quality. In recent years, the rapid advancement of Unmanned Aerial Vehicle (UAV) and sensor technologies enabled high-resolution spatial, spectral, and temporal data collection with a lower cost. Coupled with cutting-edge artificial intelligence and deep learning (AI/DL) algorithms, UAV remote sensing has become an important tool in a variety of agricultural applications. This study aims to investigate the potential of UAV-based multitemporal multispectral data for preharvest wheat yield, test weight and protein content estimation under the framework of AI/DL. UAV-based multispectral images were collected throughout the 2022 winter wheat growing season over seven experimental winter wheat fields across South Dakota, USA. Plot-level canopy spectral and texture features were derived from UAV multispectral imagery. Traditional machine learning approaches such as Partial Least Squares Regression, Support Vector Regression, and Random Forest Regression were employed to develop prediction models using plot-level averaged spectral and texture features. Additionally, deep learning methods such as Convolutional Neural Networks (CNN) and hybrid CNN and Long-Short Term Memory (CNN-LSTM) were also implemented using plot-level reflectance imagery as input for prediction model development. This research highlights the potential of coupling high-resolution UAV remote sensing with cutting-edge AI/DL in predicting wheat yield, test wight and protein content. The results from this work deliver valuable insights for high-throughput phenotyping and crop field management with high spatial precision.

Key words: artificial intelligence (AI), deep learning (DL), unmanned aerial vehicle (UAV), remote sensing, winter wheat, yield prediction