A Deep Learning–Based Method for In-Field Measuring Flag Leaf Angle for Wheat

A research team led by Prof. JIANG Ni from the Institute of Genetics and Developmental Biology of the Chinese Academy of Sciences, proposed a cost-effective method for in-field acquisition of flag leaf angle (FLANG) images and developed a lightweight deep learning model, LeafPoseNet, for accurate FLANG estimation.

 

The findings were published on July 25 in The Crop Journal (h ttps://doi.org/10.1016/j.cj.2025.07.002).

 

FLANG is one of the key traits in wheat breeding due to its important role in plant architecture, light interception, and yield potential. Currently, its measurement remains a bottleneck in high-throughput phenotyping for modern wheat breeding and management due to its reliance on labor-intensive and subjective manual measurement.

 

To address this challenge, the team developed LeafPoseNet, a keypoint-based pose detection model that automatically detects three keypoints: the center of the flag leaf (Point L), the junction between the flag leaf and stem (Point J), and the center of the stem (Point S), thereby enabling automated calculation of FLANG.

 

Compared with state-of-the-art keypoint detection models, LeafPoseNet demonstrates superior performance, achieving a mean absolute error (MAE) of 1.75°, a root mean square error (RMSE) of 2.17°, and a coefficient of determination (R²) of 0.998. It exhibits robust keypoint localization across diverse leaf shapes and complex scenarios. With its lightweight architecture and high computational efficiency, the model is readily deployable on smartphones, enabling high-throughput, rapid in-field measurements.

 

The team demonstrated the application of LeafPoseNet to measure FLANG in a diverse panel of 221 bread wheat accessions. Using a mixed linear model (MLM) for genome-wide association study (GWAS), they identified 10 quantitative trait loci (QTL) associated with FLANG, providing important insights into genetic architecture of FLANG in wheat.

 

This study provides a handy tool for high-throughput in-field measurement of FLANG in wheat, which facilitates breeding and genetic analysis.

 

This work was supported by the Biological Breeding-National Science and Technology Major Project, the National Key Research and Development Program of China, and the Strategic Priority Research Program of the Chinese Academy of Sciences.

 

 

Contact:

Prof. JIANG Ni

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences

Email: njiang@genetics.ac.cn