Scholarlink

<p>Plant diseases severely impact crop quality and yield, necessitating early and precise detection to minimize economic losses and prevent outbreaks. This research introduces a&#xa0;framework for disease identification and classification in fruit and vegetables, leveraging the advanced You Only Look Once-V5 (YOLO-V5) object detection network. The framework utilizes the enhanced PlantVillage dataset, which has undergone class alignment, data augmentation, and meticulous annotation to improve model accuracy and generalization. Five YOLO-V5 variations: n, s, m, l, and x were evaluated, with YOLO-V5x demonstrating the best performance. It achieved 98.5% precision, 96.4% recall, 97.6% F1-score, and 88.1% mean average precision (mAP) (0.5–0.95 IoU). These results confirm YOLO-V5x’s effectiveness for reliable leaf disease detection in agricultural settings.</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

YOLOv5 Revisited: A Lightweight yet Accurate Framework for Plant Disease Detection in Agricultural Applications

  • 期刊论文

  • 发表日期：

    2025年11月6日

查看全文
  • Huria Ali,
  • Muhammad Imran,
  • Saad Irfan Khan,
  • Anees Tariq,
  • Hassan Dawood,
  • Hussain Dawood

摘要

Plant diseases severely impact crop quality and yield, necessitating early and precise detection to minimize economic losses and prevent outbreaks. This research introduces a framework for disease identification and classification in fruit and vegetables, leveraging the advanced You Only Look Once-V5 (YOLO-V5) object detection network. The framework utilizes the enhanced PlantVillage dataset, which has undergone class alignment, data augmentation, and meticulous annotation to improve model accuracy and generalization. Five YOLO-V5 variations: n, s, m, l, and x were evaluated, with YOLO-V5x demonstrating the best performance. It achieved 98.5% precision, 96.4% recall, 97.6% F1-score, and 88.1% mean average precision (mAP) (0.5–0.95 IoU). These results confirm YOLO-V5x’s effectiveness for reliable leaf disease detection in agricultural settings.