Scholarlink

<p>Agricultural production is crucial for global economies, yet crop diseases significantly threaten productivity and food security. Traditional manual diagnosis methods are labor-intensive and subjective, leading to inaccurate pesticide applications. Leveraging deep learning, recent advancements have improved disease identification accuracy. However, existing models lack generalization across multiple crops and vegetable diseases. To address this gap, a comprehensive dataset comprising 11 common leaf diseases, such as Anthracnose, Bacterial Blight, Bacterial Spot, Bacterial Wilt, Blast, Downy Mildew, Early Blight, Late Blight, Mosaic, Powdery Mildew, and Rust, affecting 20 widely consumable crops and vegetables, such as rice, wheat, corn, tea, coffee, soybean, potato, tomato, carrot, black gram, pea, cassava, sugarcane, bottle gourd, pepper bell, brinjal, lettuce, cabbage, cauliflower, and cucumber, was curated. Data augmentation techniques expanded the dataset to 16,800 images to enhance the robustness by reducing the overfitting of the deep learning disease recognition model. The dataset is publicly available on GitHub for research purposes. In addition to dataset preparation, this paper introduces a generalized deep learning model for efficient recognition of the leaf diseases of crops and vegetables. Emphasis is placed on optimizing hyperparameters for well-known pre-trained deep learning models, including DenseNet-121, ResNet-50, VGG-16, VGG-19, and Inception-V4. Experimental results using the dataset demonstrate that DenseNet-121 achieved a classification accuracy of 97.58%, surpassing the other models.</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

A generalized leaf disease recognition of various crops and vegetables through computer vision and machine learning

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • Nusrat Sultana,
  • Sabrina Sharmin,
  • Mohammad Shorif Uddin

摘要

Agricultural production is crucial for global economies, yet crop diseases significantly threaten productivity and food security. Traditional manual diagnosis methods are labor-intensive and subjective, leading to inaccurate pesticide applications. Leveraging deep learning, recent advancements have improved disease identification accuracy. However, existing models lack generalization across multiple crops and vegetable diseases. To address this gap, a comprehensive dataset comprising 11 common leaf diseases, such as Anthracnose, Bacterial Blight, Bacterial Spot, Bacterial Wilt, Blast, Downy Mildew, Early Blight, Late Blight, Mosaic, Powdery Mildew, and Rust, affecting 20 widely consumable crops and vegetables, such as rice, wheat, corn, tea, coffee, soybean, potato, tomato, carrot, black gram, pea, cassava, sugarcane, bottle gourd, pepper bell, brinjal, lettuce, cabbage, cauliflower, and cucumber, was curated. Data augmentation techniques expanded the dataset to 16,800 images to enhance the robustness by reducing the overfitting of the deep learning disease recognition model. The dataset is publicly available on GitHub for research purposes. In addition to dataset preparation, this paper introduces a generalized deep learning model for efficient recognition of the leaf diseases of crops and vegetables. Emphasis is placed on optimizing hyperparameters for well-known pre-trained deep learning models, including DenseNet-121, ResNet-50, VGG-16, VGG-19, and Inception-V4. Experimental results using the dataset demonstrate that DenseNet-121 achieved a classification accuracy of 97.58%, surpassing the other models.