Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations

Above-ground biomass (AGB) is a critical phenotype representing crop growth. Non-invasive evaluations of AGB, including deep-learning-based red-green-blue (RGB) image analyses, are often specific to the training data. The robustness of the estimation model across untrained conditions is essential to...

Full description

Bibliographic Details
Main Authors: Nakajima, Kota, Saito, Kazuki, Tsujimoto, Yasuhiro, Takai, Toshiyuki, Mochizuki, Atsushi, Yamaguchi, Tomoaki, Ibrahim, Ali, Mairoua, Salifou Goube, Andrianary, Bruce Haja, Katsura, Keisuke, Tanaka, Yu
Format: Journal Article
Language:Inglés
Published: Elsevier 2025
Subjects:
Online Access:https://hdl.handle.net/10568/175950
_version_ 1855530161452089344
author Nakajima, Kota
Saito, Kazuki
Tsujimoto, Yasuhiro
Takai, Toshiyuki
Mochizuki, Atsushi
Yamaguchi, Tomoaki
Ibrahim, Ali
Mairoua, Salifou Goube
Andrianary, Bruce Haja
Katsura, Keisuke
Tanaka, Yu
author_browse Andrianary, Bruce Haja
Ibrahim, Ali
Katsura, Keisuke
Mairoua, Salifou Goube
Mochizuki, Atsushi
Nakajima, Kota
Saito, Kazuki
Takai, Toshiyuki
Tanaka, Yu
Tsujimoto, Yasuhiro
Yamaguchi, Tomoaki
author_facet Nakajima, Kota
Saito, Kazuki
Tsujimoto, Yasuhiro
Takai, Toshiyuki
Mochizuki, Atsushi
Yamaguchi, Tomoaki
Ibrahim, Ali
Mairoua, Salifou Goube
Andrianary, Bruce Haja
Katsura, Keisuke
Tanaka, Yu
author_sort Nakajima, Kota
collection Repository of Agricultural Research Outputs (CGSpace)
description Above-ground biomass (AGB) is a critical phenotype representing crop growth. Non-invasive evaluations of AGB, including deep-learning-based red-green-blue (RGB) image analyses, are often specific to the training data. The robustness of the estimation model across untrained conditions is essential to monitor crop productivity globally, but it has yet to be fully assessed. This study aims to assess the robustness of a convolutional neural network (CNN) model for rice AGB estimation across five locations in three countries, and to demonstrate the feasibility of robust model via a practical approach. From transplanting to heading, 1957 RGB images were captured vertically downward over the rice canopy, covering approximately 1 m2. First, a base model was established using data collected from a single location. Then, its robustness was assessed using test datasets taken from the other four locations. The CNN model showed a significant variation in estimation accuracy across the untrained four locations, indicating insufficient robustness of the base model. Subsequently, we quantitatively tested the impact of improving training data diversity on model robustness by adding data from each of the four locations to the base model's training data. Adding at most 48 data points from a location achieved practical accuracy for the added location, with R2Adabove 0.8. Interestingly, adding data from one location sometimes improved the accuracy for other untrained locations as well. These findings suggest that collecting diverse training data for RGB-based estimation, combined with evaluation of robustness paves the way for on-site and instant AGB monitoring of rice.
format Journal Article
id CGSpace175950
institution CGIAR Consortium
language Inglés
publishDate 2025
publishDateRange 2025
publishDateSort 2025
publisher Elsevier
publisherStr Elsevier
record_format dspace
spelling CGSpace1759502025-11-12T04:58:14Z Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations Nakajima, Kota Saito, Kazuki Tsujimoto, Yasuhiro Takai, Toshiyuki Mochizuki, Atsushi Yamaguchi, Tomoaki Ibrahim, Ali Mairoua, Salifou Goube Andrianary, Bruce Haja Katsura, Keisuke Tanaka, Yu above ground biomass crop growth crop monitoring data collection field experiments transplanting machine learning remote sensing Above-ground biomass (AGB) is a critical phenotype representing crop growth. Non-invasive evaluations of AGB, including deep-learning-based red-green-blue (RGB) image analyses, are often specific to the training data. The robustness of the estimation model across untrained conditions is essential to monitor crop productivity globally, but it has yet to be fully assessed. This study aims to assess the robustness of a convolutional neural network (CNN) model for rice AGB estimation across five locations in three countries, and to demonstrate the feasibility of robust model via a practical approach. From transplanting to heading, 1957 RGB images were captured vertically downward over the rice canopy, covering approximately 1 m2. First, a base model was established using data collected from a single location. Then, its robustness was assessed using test datasets taken from the other four locations. The CNN model showed a significant variation in estimation accuracy across the untrained four locations, indicating insufficient robustness of the base model. Subsequently, we quantitatively tested the impact of improving training data diversity on model robustness by adding data from each of the four locations to the base model's training data. Adding at most 48 data points from a location achieved practical accuracy for the added location, with R2Adabove 0.8. Interestingly, adding data from one location sometimes improved the accuracy for other untrained locations as well. These findings suggest that collecting diverse training data for RGB-based estimation, combined with evaluation of robustness paves the way for on-site and instant AGB monitoring of rice. 2025-08 2025-08-04T03:02:18Z 2025-08-04T03:02:18Z Journal Article https://hdl.handle.net/10568/175950 en Open Access application/pdf Elsevier Nakajima, Kota, Kazuki Saito, Yasuhiro Tsujimoto, Toshiyuki Takai, Atsushi Mochizuki, Tomoaki Yamaguchi, Ali Ibrahim et al. "Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations." Smart Agricultural Technology 11 (2025): 100998.
spellingShingle above ground biomass
crop growth
crop monitoring
data collection
field experiments
transplanting
machine learning
remote sensing
Nakajima, Kota
Saito, Kazuki
Tsujimoto, Yasuhiro
Takai, Toshiyuki
Mochizuki, Atsushi
Yamaguchi, Tomoaki
Ibrahim, Ali
Mairoua, Salifou Goube
Andrianary, Bruce Haja
Katsura, Keisuke
Tanaka, Yu
Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title_full Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title_fullStr Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title_full_unstemmed Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title_short Robustness of the RGB image-based estimation for rice above-ground biomass by utilizing the dataset collected across multiple locations
title_sort robustness of the rgb image based estimation for rice above ground biomass by utilizing the dataset collected across multiple locations
topic above ground biomass
crop growth
crop monitoring
data collection
field experiments
transplanting
machine learning
remote sensing
url https://hdl.handle.net/10568/175950
work_keys_str_mv AT nakajimakota robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT saitokazuki robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT tsujimotoyasuhiro robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT takaitoshiyuki robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT mochizukiatsushi robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT yamaguchitomoaki robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT ibrahimali robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT mairouasalifougoube robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT andrianarybrucehaja robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT katsurakeisuke robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations
AT tanakayu robustnessofthergbimagebasedestimationforriceabovegroundbiomassbyutilizingthedatasetcollectedacrossmultiplelocations