Scholarlink

<p>Timely durian leaf disease detection is critical for Vietnam’s agricultural productivity, yet traditional methods remain labor-intensive and error-prone. This study proposes a&#xa0;hybrid pipeline integrating deep transfer learning with binary Particle Swarm Optimization (PSO) for efficient disease classification. Three lightweight backbones, such as MobileNetV3-Large, EfficientNet-B0, and EfficientNetV2-B0 to extract 128-dimensional features from a&#xa0;real-world Vietnamese durian dataset (2595 images, 6&#xa0;classes), which PSO then prunes using five-fold support vector machine (SVM) cross-validation fitness. The optimized subsets were evaluated across five machine learning (ML) classifiers, achieving up to 92.6% test accuracy with a&#xa0;modest improvement over the baseline while reducing dimensionality. PSO-selected features demonstrated the potential for accelerated inference and interpretable agricultural diagnostics on resource-constrained devices.</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

Transfer Learning with Particle Swarm Optimization for Durian Leaf Disease Image Classification

  • 期刊论文

  • 发表日期：

    2026年4月27日

查看全文
  • Tran Nguyen Phi Hung,
  • Nguyen Minh Tuan

摘要

Timely durian leaf disease detection is critical for Vietnam’s agricultural productivity, yet traditional methods remain labor-intensive and error-prone. This study proposes a hybrid pipeline integrating deep transfer learning with binary Particle Swarm Optimization (PSO) for efficient disease classification. Three lightweight backbones, such as MobileNetV3-Large, EfficientNet-B0, and EfficientNetV2-B0 to extract 128-dimensional features from a real-world Vietnamese durian dataset (2595 images, 6 classes), which PSO then prunes using five-fold support vector machine (SVM) cross-validation fitness. The optimized subsets were evaluated across five machine learning (ML) classifiers, achieving up to 92.6% test accuracy with a modest improvement over the baseline while reducing dimensionality. PSO-selected features demonstrated the potential for accelerated inference and interpretable agricultural diagnostics on resource-constrained devices.