Scholarlink

This paper presents an optimized deep learning framework for fruit disease detection, focusing on apple and grape leaves. The study leverages the Inception-ResNet-V2 model, pre-trained on ImageNet, for feature extraction due to its ability to capture multi-scale patterns crucial for detecting complex disease symptoms. Several neural network architectures, including trilayered, bilayered, medium, and wide models, were evaluated for performance. The wide neural network achieved the highest accuracy at 98.5%, outperforming Inception-ResNet-V2’s 90.1%. Preprocessing techniques such as contrast enhancement, data augmentation, and entropy-based feature selection were employed to improve both classification accuracy and computational efficiency. This framework, integrating feature fusion and deep learning, demonstrates significant potential for enhancing fruit disease detection accuracy, contributing to precision agriculture by automating disease management.

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

Optimized Deep Learning Framework for Fruit Disease Detection Using Feature Fusion and Neural Network Architectures

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • K. V. Narasimha Reddy,
  • Shaik Jaleel,
  • Puli Sailokesh Reddy,
  • Tirupathi Eswar Vara Prasad,
  • Vemula Rajesh,
  • Dodda Venkata Reddy,
  • Sireesha Moturi

摘要

This paper presents an optimized deep learning framework for fruit disease detection, focusing on apple and grape leaves. The study leverages the Inception-ResNet-V2 model, pre-trained on ImageNet, for feature extraction due to its ability to capture multi-scale patterns crucial for detecting complex disease symptoms. Several neural network architectures, including trilayered, bilayered, medium, and wide models, were evaluated for performance. The wide neural network achieved the highest accuracy at 98.5%, outperforming Inception-ResNet-V2’s 90.1%. Preprocessing techniques such as contrast enhancement, data augmentation, and entropy-based feature selection were employed to improve both classification accuracy and computational efficiency. This framework, integrating feature fusion and deep learning, demonstrates significant potential for enhancing fruit disease detection accuracy, contributing to precision agriculture by automating disease management.