Scholarlink

<p>Plant disease detection is particularly a critical complexity in the agriculture farming field. Earlier and precise detection of leaf disease in plants aids in enhancing crop production and economy while reducing environmental damage. Disease prognosis in plants requires more work, and information on plant diseases. Therefore, plant disease detection is the major potential technique in agriculture that is gaining essential interest in farming as well as computer communities. To subdue this issue, a systematic framework is proposed for multi-class leaf disease detection in the Internet of Things (IoT) based sustainable agriculture using the Deep Spiking Kronecker Network (DSKN). The IoT nodes are simulated and then routing is performed by the Multi-objective Fractional Artificial Bee Colony Algorithm algorithm. At Base Station, multi-class leaf disease detection is executed. The image is pre-processed by a Gaussian filter. The crop leaf disease area segmentation is carried out by UNeXt and then image augmentation is done using position and color augmentation. Afterwards, feature extraction is performed. Thereafter, leaf-type classification is accomplished by the proposed DSKN. After the identification of type, the detection of multi-class crop leaf disease is achieved using the proposed DSKN. The performance of the proposed DSKN is evaluated using accuracy, True Positive rate (TPR) and False Positive rate (FPR) metrics based on plant village database (<a href="https://github.com/spMohanty/PlantVillage-Dataset/tree/master/raw/color">https://github.com/spMohanty/PlantVillage-Dataset/tree/master/raw/color</a>) and Crop Disease Image Dataset (<a href="https://www.kaggle.com/datasets/jawadali1045/20k-multi-class-crop-disease-images">https://www.kaggle.com/datasets/jawadali1045/20k-multi-class-crop-disease-images</a>). It observed that the proposed method obtained best outcomes of 90.76%, 90.75% and 9.35% for accuracy, TPR and FPR using the plant village database.</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

DSKN: Deep Spiking Kronecker Network for leaf type classification and multi-class leaf disease detection in internet of things based sustainable agriculture

  • 期刊论文

  • 发表日期：

    2025年5月31日

查看全文
  • Nandkumar Prabhakar Kulkarni,
  • Bhuvaneshwari Jolad,
  • Amol Govind Patil

摘要

Plant disease detection is particularly a critical complexity in the agriculture farming field. Earlier and precise detection of leaf disease in plants aids in enhancing crop production and economy while reducing environmental damage. Disease prognosis in plants requires more work, and information on plant diseases. Therefore, plant disease detection is the major potential technique in agriculture that is gaining essential interest in farming as well as computer communities. To subdue this issue, a systematic framework is proposed for multi-class leaf disease detection in the Internet of Things (IoT) based sustainable agriculture using the Deep Spiking Kronecker Network (DSKN). The IoT nodes are simulated and then routing is performed by the Multi-objective Fractional Artificial Bee Colony Algorithm algorithm. At Base Station, multi-class leaf disease detection is executed. The image is pre-processed by a Gaussian filter. The crop leaf disease area segmentation is carried out by UNeXt and then image augmentation is done using position and color augmentation. Afterwards, feature extraction is performed. Thereafter, leaf-type classification is accomplished by the proposed DSKN. After the identification of type, the detection of multi-class crop leaf disease is achieved using the proposed DSKN. The performance of the proposed DSKN is evaluated using accuracy, True Positive rate (TPR) and False Positive rate (FPR) metrics based on plant village database (https://github.com/spMohanty/PlantVillage-Dataset/tree/master/raw/color) and Crop Disease Image Dataset (https://www.kaggle.com/datasets/jawadali1045/20k-multi-class-crop-disease-images). It observed that the proposed method obtained best outcomes of 90.76%, 90.75% and 9.35% for accuracy, TPR and FPR using the plant village database.