Scholarlink

<p>The citrus longhorn beetle (<i>Anoplophora chinensis </i>Forster) is a&#xa0;highly destructive pest that can cause tree mortality in many hosts, including stone fruit and forest tree species. This pest is on the A2 list of the European and Mediterranean Plant Protection Organization (EPPO) and is kept under quarantine in Türkiye. This study focused on detecting <i>A.&#xa0;chinensis</i>, a&#xa0;significant pest responsible for substantial yield and quality reductions in hazelnut (<i>Corylus avellana</i>) production, using field-based image analysis techniques. Adult beetles were collected from three hazelnut orchards in Serdivan, Sakarya, and identified based on morphological characteristics. High-resolution images were captured both in the field and laboratory, then labeled using MakeSense software. A&#xa0;model was developed using the YOLOv5 (You Only Look Once, Version&#xa0;5) deep learning framework in Python. The analysis was conducted over 100 epochs, using 87 images and a&#xa0;total of 157 layers for feature extraction. The dataset was divided into training (70%), validation (15%), and testing (15%) subsets. The final model demonstrated a&#xa0;99% accuracy rate in species identification. The results highlight the efficacy of deep learning-based image analysis for early detection and monitoring of <i>A.&#xa0;chinensis</i>. This study contributes to accurate species identification providing valuable insights for pest management strategies.</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

Detection and Quantification of Citrus Long-horned Beetle (Anoplophora chinensis) in Hazelnut Orchards Using YOLOv5 Deep Learning

  • 期刊论文

  • 发表日期：

    2025年7月7日

查看全文
  • Bahadır Şin,
  • Lerzan Öztürk

摘要

The citrus longhorn beetle (Anoplophora chinensis Forster) is a highly destructive pest that can cause tree mortality in many hosts, including stone fruit and forest tree species. This pest is on the A2 list of the European and Mediterranean Plant Protection Organization (EPPO) and is kept under quarantine in Türkiye. This study focused on detecting A. chinensis, a significant pest responsible for substantial yield and quality reductions in hazelnut (Corylus avellana) production, using field-based image analysis techniques. Adult beetles were collected from three hazelnut orchards in Serdivan, Sakarya, and identified based on morphological characteristics. High-resolution images were captured both in the field and laboratory, then labeled using MakeSense software. A model was developed using the YOLOv5 (You Only Look Once, Version 5) deep learning framework in Python. The analysis was conducted over 100 epochs, using 87 images and a total of 157 layers for feature extraction. The dataset was divided into training (70%), validation (15%), and testing (15%) subsets. The final model demonstrated a 99% accuracy rate in species identification. The results highlight the efficacy of deep learning-based image analysis for early detection and monitoring of A. chinensis. This study contributes to accurate species identification providing valuable insights for pest management strategies.