Related Experiment Video
Updated: Jan 8, 2026
11:49
Cereal Crop Ear Counting in Field Conditions Using Zenithal RGB Images
Published on: February 2, 2019
9.8K
RGB imaging and computer vision-based approaches for identifying spike number loci for wheat
Lei Li1,2,3, Muhammad Adeel Hassan4,5, Duoxia Wang1
1State Key Laboratory of Crop Gene Resources and Breeding, Institute of Crop Sciences, National Wheat Improvement Centre, Chinese Academy of Agricultural Sciences (CAAS), Beijing, 100081, China.
Plant Phenomics (Washington, D.C.)
|December 19, 2025
Summary
Developing efficient wheat spike counting methods is crucial for breeding. A new YOLOX-P algorithm improves accuracy in detecting spike number (SN) and identifies new genetic loci for wheat yield improvement.
Area of Science:
- Agricultural Science
- Plant Breeding
- Computational Biology
Background:
- Spike number (SN) is a key determinant of wheat grain yield.
- Manual SN counting is labor-intensive and limits large-scale breeding programs.
- Accurate and efficient SN quantification is essential for advancing wheat genetics and breeding.
Purpose of the Study:
- To optimize wheat spike detection models for high-throughput phenotyping.
- To develop an improved YOLOX-P algorithm for enhanced SN identification accuracy.
- To identify novel quantitative trait loci (QTLs) associated with SN in wheat.
Main Methods:
- Utilized the YOLOX algorithm to evaluate optimal growth stages for SN detection in wheat.
- Developed an enhanced YOLOX-P algorithm incorporating attention modules and increased image resolution.
- Employed 50k SNP arrays for genome-wide association studies to identify SN loci.
- Developed kompetitive allele-specific PCR markers for validated loci.
Main Results:
- The late grain-filling stage demonstrated the highest SN detection accuracy (91.8-95.02%).
- The YOLOX-P algorithm significantly outperformed the standard YOLOX algorithm in precision and F1 scores.
- Identified three novel SN loci: QSN.caas-4A2, QSN.caas-4D, and QSN.caas-5B2.
- Validated the genetic effects of two loci (QSN.caas-4A2 and QSN.caas-5B2) in a diverse cultivar panel.
Conclusions:
- The late grain-filling stage is optimal for automated wheat SN counting.
- The YOLOX-P algorithm offers a robust tool for high-throughput SN phenotyping.
- The identified SN loci provide valuable targets for marker-assisted selection in wheat breeding.
- This research facilitates efficient genetic improvement of wheat yield potential.
