High-throughput fruit-level phenotyping captures quantitative variation in Botrytis cinerea resistance across strawberry breeding populations

Background Accurate and fast screening of Botrytis cinerea lesions on strawberry fruits is vital for breeding programs focused on developing disease-resistant cultivars. Traditional manual inspection is labor-intensive and subjective, while existing automated methods often require extensive labeled data and struggle with variability in lesion appearance. Results Smartphone images of individual strawberry fruits arranged on trays were collected during disease trials, representing 350 distinct strawberry accessions. The images were analyzed using a multi-model deep learning algorithm that integrates fine-tuned task-specific models with text prompt-based models, generating key phenotypic metrics including strawberry fruit size, Botrytis lesion size, and Botrytis lesion to strawberry fruit size ratio – parameters that are difficult to quantify manually. The text prompt-based models were used for improving the generalizability of the over-all algorithm by supporting the pre-annotation of training datasets and providing complementary detection to the fine-tuned models. The algorithm achieved high performances on strawberry fruit instance segmentation (F 1 -score = 0.97; mIoU seg = 0.92) and consistent Botrytis lesion segmentation (mIoU seg = 0.97) across diverse strawberry accessions. Comparative analysis between subjective and objective data indicated that objective measurements provide higher discriminatory ability, highlighting the utility of automated phenotyping in breeding programs. Conclusion To the best of our knowledge, this is the first work to develop a fruit-level multi-model deep learning approach for automated Botrytis lesion screening, facilitating high-throughput phenotyping in breeding trials. This work presents a practical, consistent, and scalable solution for plant disease screening that can be implemented using accessible imaging devices and moderate computational resources, reducing barriers to the adoption of AI-based phenotyping approaches in breeding programs. This work may support the development of more efficient marker-assisted and genomic selection strategies in breeding programs.

Authors

Publication Details

Journal
Plant Methods
Published
2026-11-19
DOI
https://doi.org/10.1186/s13007-026-01616-4
Primary Topic
Smart Agriculture and AI
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

High-throughput fruit-level phenotyping captures quantitative variation in Botrytis cinerea resistance across strawberry breeding populations

Richard G. F. Visser, Manon Mensink, Suzan H. E. J. Gabriëls, Paul Arens et al.
Plant Methods
Smart Agriculture and AI
article

High-throughput fruit-level phenotyping captures quantitative variation in Botrytis cinerea resistance across strawberry breeding populations

Richard G. F. Visser, Manon Mensink, Suzan H. E. J. Gabriëls, Paul Arens, Dan Jeric Arcega Rustia, Maikel Zerdoner
article en

Abstract

Background Accurate and fast screening of Botrytis cinerea lesions on strawberry fruits is vital for breeding programs focused on developing disease-resistant cultivars. Traditional manual inspection is labor-intensive and subjective, while existing automated methods often require extensive labeled data and struggle with variability in lesion appearance. Results Smartphone images of individual strawberry fruits arranged on trays were collected during disease trials, representing 350 distinct strawberry accessions. The images were analyzed using a multi-model deep learning algorithm that integrates fine-tuned task-specific models with text prompt-based models, generating key phenotypic metrics including strawberry fruit size, Botrytis lesion size, and Botrytis lesion to strawberry fruit size ratio – parameters that are difficult to quantify manually. The text prompt-based models were used for improving the generalizability of the over-all algorithm by supporting the pre-annotation of training datasets and providing complementary detection to the fine-tuned models. The algorithm achieved high performances on strawberry fruit instance segmentation (F 1 -score = 0.97; mIoU seg = 0.92) and consistent Botrytis lesion segmentation (mIoU seg = 0.97) across diverse strawberry accessions. Comparative analysis between subjective and objective data indicated that objective measurements provide higher discriminatory ability, highlighting the utility of automated phenotyping in breeding programs. Conclusion To the best of our knowledge, this is the first work to develop a fruit-level multi-model deep learning approach for automated Botrytis lesion screening, facilitating high-throughput phenotyping in breeding trials. This work presents a practical, consistent, and scalable solution for plant disease screening that can be implemented using accessible imaging devices and moderate computational resources, reducing barriers to the adoption of AI-based phenotyping approaches in breeding programs. This work may support the development of more efficient marker-assisted and genomic selection strategies in breeding programs.

Plant Methods
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