Standardized Image-Based Phenotyping of Tomato Bacterial Wilt Symptom Severity using Deep Learning to Support Resistance Screening

Bacterial wilt, caused by the soil-borne bacterium Ralstonia pseudosolanacearum, severely limits tomato production, making objective and reproducible disease phenotyping essential for resistance breeding.In this study, a convolutional neural network (CNN) image classification model based on transfer learning with EfficientNetB0 was developed to classify tomato bacterial wilt severity under controlled inoculation and imaging conditions to support standardized resistance screening.Four susceptible tomato cultivars were inoculated with R. pseudosolanacearum strain SL341 using drench and petiole methods to generate diverse wilt symptoms.A total of 1,500 standardized multi-view images from 776 individual plants were categorized into low-severity, intermediateseverity, and severe-wilt classes.To prevent data leakage, images were partitioned at the plant level into training, validation, and independent test sets at an approximate ratio of 64:16:20.On-the-fly image augmentation was applied during training to improve generalization.The final model correctly classified 288 of 300 test images, achieving 96.0% overall accuracy.Performance was highest for severe symptoms and lowest for the intermediate class, indicating greater difficulty in distinguishing transitional disease stages.A Grad-CAM analysis showed that predictions were based on biologically relevant plant regions rather than background features.Training and validation curves also showed stable convergence without substantial overfitting.These results demonstrate that deep-learning-assisted image classification can provide an objective, reproducible auxiliary tool for standardized symptom scoring, thereby relieving the phenotyping bottleneck in tomato bacterial wilt resistance screening.

Authors

Institutions

Publication Details

Journal
Horticultural Science and Technology
Published
2026-11-13
DOI
https://doi.org/10.7235/hort.20260049
Primary Topic
Smart Agriculture and AI
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Standardized Image-Based Phenotyping of Tomato Bacterial Wilt Symptom Severity using Deep Learning to Support Resistance Screening

Ji-Su Kwon, June Sung Lee, Seon‐In Yeom, You-Jin Lee
1 links
Horticultural Science and Technology
Smart Agriculture and AI
article

Standardized Image-Based Phenotyping of Tomato Bacterial Wilt Symptom Severity using Deep Learning to Support Resistance Screening

Ji-Su Kwon, June Sung Lee, Seon‐In Yeom, You-Jin Lee
article en

Abstract

Bacterial wilt, caused by the soil-borne bacterium Ralstonia pseudosolanacearum, severely limits tomato production, making objective and reproducible disease phenotyping essential for resistance breeding.In this study, a convolutional neural network (CNN) image classification model based on transfer learning with EfficientNetB0 was developed to classify tomato bacterial wilt severity under controlled inoculation and imaging conditions to support standardized resistance screening.Four susceptible tomato cultivars were inoculated with R. pseudosolanacearum strain SL341 using drench and petiole methods to generate diverse wilt symptoms.A total of 1,500 standardized multi-view images from 776 individual plants were categorized into low-severity, intermediateseverity, and severe-wilt classes.To prevent data leakage, images were partitioned at the plant level into training, validation, and independent test sets at an approximate ratio of 64:16:20.On-the-fly image augmentation was applied during training to improve generalization.The final model correctly classified 288 of 300 test images, achieving 96.0% overall accuracy.Performance was highest for severe symptoms and lowest for the intermediate class, indicating greater difficulty in distinguishing transitional disease stages.A Grad-CAM analysis showed that predictions were based on biologically relevant plant regions rather than background features.Training and validation curves also showed stable convergence without substantial overfitting.These results demonstrate that deep-learning-assisted image classification can provide an objective, reproducible auxiliary tool for standardized symptom scoring, thereby relieving the phenotyping bottleneck in tomato bacterial wilt resistance screening.

Horticultural Science and TechnologyVol. 44
Gyeongsang National University (KR)
National Research Foundation of Korea
Zero hunger
Smart Agriculture and AI
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon