Hongkui Zhou , Fudeng Huang , Weidong Lou , Qing Gu , Ziran Ye , Hao Hu , Xiaobin Zhang
{"title":"Yield prediction through UAV-based multispectral imaging and deep learning in rice breeding trials","authors":"Hongkui Zhou , Fudeng Huang , Weidong Lou , Qing Gu , Ziran Ye , Hao Hu , Xiaobin Zhang","doi":"10.1016/j.agsy.2024.104214","DOIUrl":null,"url":null,"abstract":"<div><h3>Context</h3><div>Predicting crop yields with high precision and timeliness is essential for crop breeding, enabling the optimization of planting strategies and efficients resource allocation while ensuring food security. Current research in this field typically does not address the problem of yield prediction in the diverse context of breeding experiments involving numerous varieties. However, evaluating the performance of prediction models across multiple varieties is vital for further model refining and enhancing model robustness and adaptability.</div></div><div><h3>Objective</h3><div>This study aims to evaluate the performance of feature- and image-based yield prediction models for yields with multiple varieties to compare their capabilities and determine an appropriate timing for early yield prediction.</div></div><div><h3>Methods</h3><div>This study combines unmanned aerial vehicle (UAV)-based multispectral remote sensing imagery with machine learning and deep learning-based algorithms to develop rice yield prediction models across multiple varieties. The performances of both feature- and image-based models are evaluated. The feature-based models considered in this study include random forest (RF), deep neural network (DNN), and long short-term memory (LSTM) algorithms, and the image-based models are convolutional neural network (CNN) architectures, including both two-dimensional (2D) and three-dimensional (3D) CNN models. To assess the performance of the multi-variety crop yield prediction models thoroughly, this study considers two sampling scenarios: stratified sampling and group sampling.</div></div><div><h3>Results and conclusions</h3><div>The results show that the image-based deep learning models outperform the feature-based machine learning models, which indicates their superior robustness in multi-variety scenarios and highlights their significant potential of directly extracting spatiotemporal features from images for yield prediction. The results indicate that the multi-temporal 2D CNN model (i.e., the CNN-M2D model) can achieve the best yield prediction performance among all models, achieving RRMSE = 8.13 % and R<sup>2</sup> = 0.73. The prediction results also demonstrate good consistency with the observed data, indicating an efficient capturing of spatial pattern variations in yield across different varieties. Based on the results, with the crops progressing along the growth stages, the accuracy of the yield prediction models improves gradually, achieving the best prediction performance during the flowering to grain-filling stage. Finally, according to the results, the optimal lead time for predicting rice yield is approximately one month before harvest.</div></div><div><h3>Significance</h3><div>Our study can provide a reference for the research community in yield prediction and high-yield variety selection in breeding trials.</div></div>","PeriodicalId":7730,"journal":{"name":"Agricultural Systems","volume":"223 ","pages":"Article 104214"},"PeriodicalIF":6.1000,"publicationDate":"2024-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Agricultural Systems","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0308521X24003640","RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"AGRICULTURE, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Context
Predicting crop yields with high precision and timeliness is essential for crop breeding, enabling the optimization of planting strategies and efficients resource allocation while ensuring food security. Current research in this field typically does not address the problem of yield prediction in the diverse context of breeding experiments involving numerous varieties. However, evaluating the performance of prediction models across multiple varieties is vital for further model refining and enhancing model robustness and adaptability.
Objective
This study aims to evaluate the performance of feature- and image-based yield prediction models for yields with multiple varieties to compare their capabilities and determine an appropriate timing for early yield prediction.
Methods
This study combines unmanned aerial vehicle (UAV)-based multispectral remote sensing imagery with machine learning and deep learning-based algorithms to develop rice yield prediction models across multiple varieties. The performances of both feature- and image-based models are evaluated. The feature-based models considered in this study include random forest (RF), deep neural network (DNN), and long short-term memory (LSTM) algorithms, and the image-based models are convolutional neural network (CNN) architectures, including both two-dimensional (2D) and three-dimensional (3D) CNN models. To assess the performance of the multi-variety crop yield prediction models thoroughly, this study considers two sampling scenarios: stratified sampling and group sampling.
Results and conclusions
The results show that the image-based deep learning models outperform the feature-based machine learning models, which indicates their superior robustness in multi-variety scenarios and highlights their significant potential of directly extracting spatiotemporal features from images for yield prediction. The results indicate that the multi-temporal 2D CNN model (i.e., the CNN-M2D model) can achieve the best yield prediction performance among all models, achieving RRMSE = 8.13 % and R2 = 0.73. The prediction results also demonstrate good consistency with the observed data, indicating an efficient capturing of spatial pattern variations in yield across different varieties. Based on the results, with the crops progressing along the growth stages, the accuracy of the yield prediction models improves gradually, achieving the best prediction performance during the flowering to grain-filling stage. Finally, according to the results, the optimal lead time for predicting rice yield is approximately one month before harvest.
Significance
Our study can provide a reference for the research community in yield prediction and high-yield variety selection in breeding trials.
期刊介绍:
Agricultural Systems is an international journal that deals with interactions - among the components of agricultural systems, among hierarchical levels of agricultural systems, between agricultural and other land use systems, and between agricultural systems and their natural, social and economic environments.
The scope includes the development and application of systems analysis methodologies in the following areas:
Systems approaches in the sustainable intensification of agriculture; pathways for sustainable intensification; crop-livestock integration; farm-level resource allocation; quantification of benefits and trade-offs at farm to landscape levels; integrative, participatory and dynamic modelling approaches for qualitative and quantitative assessments of agricultural systems and decision making;
The interactions between agricultural and non-agricultural landscapes; the multiple services of agricultural systems; food security and the environment;
Global change and adaptation science; transformational adaptations as driven by changes in climate, policy, values and attitudes influencing the design of farming systems;
Development and application of farming systems design tools and methods for impact, scenario and case study analysis; managing the complexities of dynamic agricultural systems; innovation systems and multi stakeholder arrangements that support or promote change and (or) inform policy decisions.