Scholarlink

<p>Plants have a crucial role in supplying food for the global population, but their susceptibility to diseases poses a significant challenge to production. Traditional methods of monitoring leaf diseases depend on human labor and are exposed to inefficiencies. The use of artificial intelligence (AI) could identify diseases at an early stage, therefore reducing their effect and surpassing the limitations of human observation. This study proposes the implementation of hybrid intelligent model that employs convolutional neural network (CNN) to detect tomato leaf diseases. Although deep learning (DL) networks exhibit outstanding effectiveness, determining and designing the optimal model structure remains a challenging problem.&#xa0;Improving the efficiency of leaf disease identification is possible by fine-tuning CNN's hyper-parameters. A dataset of tomato leaf diseases consisting of 11,000 images, classified into 10 distinct categories has been utilized to train the proposed model. To improve the efficiency of CNN in detecting tomato plant leaf diseases, CNN’s hyper-parameters have been optimized by employing the firefly algorithm (FA). The proposed hybrid model outperforms other standard hybrid models as well as other CNN versions such as classical CNN, ResNet50, MobileNetV2, VGG16, genetic algorithm (GA) + CNN and particle swarm optimization (PSO) + CNN. The experimental analysis demonstrated that the proposed FA + CNN model has a detection accuracy of 98.96%.</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

Improved Hybrid Intelligent Framework Using Convolutional Neural Network and Firefly Optimization for Tomato Leaf Disease Detection

  • 期刊论文

  • 发表日期：

    2025年7月9日

查看全文
  • Bhabanisankar Jena,
  • Ashanta Ranjan Routray,
  • Janmenjoy Nayak,
  • Pandit Byomakesha Dash

摘要

Plants have a crucial role in supplying food for the global population, but their susceptibility to diseases poses a significant challenge to production. Traditional methods of monitoring leaf diseases depend on human labor and are exposed to inefficiencies. The use of artificial intelligence (AI) could identify diseases at an early stage, therefore reducing their effect and surpassing the limitations of human observation. This study proposes the implementation of hybrid intelligent model that employs convolutional neural network (CNN) to detect tomato leaf diseases. Although deep learning (DL) networks exhibit outstanding effectiveness, determining and designing the optimal model structure remains a challenging problem. Improving the efficiency of leaf disease identification is possible by fine-tuning CNN's hyper-parameters. A dataset of tomato leaf diseases consisting of 11,000 images, classified into 10 distinct categories has been utilized to train the proposed model. To improve the efficiency of CNN in detecting tomato plant leaf diseases, CNN’s hyper-parameters have been optimized by employing the firefly algorithm (FA). The proposed hybrid model outperforms other standard hybrid models as well as other CNN versions such as classical CNN, ResNet50, MobileNetV2, VGG16, genetic algorithm (GA) + CNN and particle swarm optimization (PSO) + CNN. The experimental analysis demonstrated that the proposed FA + CNN model has a detection accuracy of 98.96%.