Scholarlink

<p>Fusarium head blight (FHB) is one of the most destructive fungal diseases in global wheat production. Traditional methods for FHB detection face limitations such as high technical expertise requirements, limited coverage scope, and insufficient timeliness, making them inadequate for modern precision agriculture management demands. To address this challenge, this study proposes MSA-YOLO, a&#xa0;task-specific lightweight optimization framework based on YOLO11 for wheat spikelet FHB detection. MSA-YOLO integrates a&#xa0;MobileOne-based lightweight backbone, an SE-enhanced C2PSA feature recalibration module, and an Adaptive Threshold Focal Loss (ATFL) function to reduce model complexity and improve the learning of difficult diseased-spikelet samples. The dataset consists of 629 wheat spike images with spikelet-level annotations, covering multiple growth and developmental stages. Results show that MSA-YOLO reduces the parameter count from 2.58 to 1.65M and computational complexity from 6.4&#xa0;GFLOPs to 3.9&#xa0;GFLOPs. Comparative analysis with YOLOv10n, YOLOv9t, YOLOv8n, and YOLOv5n models indicates that MSA-YOLO maintains competitive detection performance with lower computational cost, supporting its potential use in lightweight wheat FHB monitoring systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

MSA-YOLO: a Lightweight Detection Model for Wheat Spikelet Fusarium Head Blight Based On YOLO 11

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Lei Shi,
  • Zilong Shang,
  • Bing Bai,
  • Yingyu Ma,
  • Fei Yin,
  • Wei Guo,
  • Yumin Chen,
  • Jingkai Lei,
  • Shufeng Xiong,
  • Yong Li

摘要

Fusarium head blight (FHB) is one of the most destructive fungal diseases in global wheat production. Traditional methods for FHB detection face limitations such as high technical expertise requirements, limited coverage scope, and insufficient timeliness, making them inadequate for modern precision agriculture management demands. To address this challenge, this study proposes MSA-YOLO, a task-specific lightweight optimization framework based on YOLO11 for wheat spikelet FHB detection. MSA-YOLO integrates a MobileOne-based lightweight backbone, an SE-enhanced C2PSA feature recalibration module, and an Adaptive Threshold Focal Loss (ATFL) function to reduce model complexity and improve the learning of difficult diseased-spikelet samples. The dataset consists of 629 wheat spike images with spikelet-level annotations, covering multiple growth and developmental stages. Results show that MSA-YOLO reduces the parameter count from 2.58 to 1.65M and computational complexity from 6.4 GFLOPs to 3.9 GFLOPs. Comparative analysis with YOLOv10n, YOLOv9t, YOLOv8n, and YOLOv5n models indicates that MSA-YOLO maintains competitive detection performance with lower computational cost, supporting its potential use in lightweight wheat FHB monitoring systems.