Scholarlink

<p>Globally, rice is important as a primary food source. However, it is susceptible to diseases that cause significant production losses. For large farmlands, manual disease surveillance is impractical and costly. To resolve this, this paper proposes a novel framework called Scalar Softmax-based Positional Encoding Deep Convolutional Neural Network (SS-PEDCNN)-centric Rice Plant Disease (RPD) Detection framework with Severity Assessment capabilities by Percentage of Infections (POI) with Fuzzy Rule. For more efficient disease detection, the proposed system considers all parts of the rice plant, encompassing the stem, sheath, and leaf. Primarily, by utilizing the Correlated Log Transformed Double Plateau Histogram Equalization (CLT-DPHE), the contrast of input images is enhanced. Afterward, by employing the Shannon Entropy-based Gaussian Mixture Model (SE-GMM), backgrounds are removed. Then, the background removed images are segmented into separate parts (stem, sheath, and leaf) utilizing Constriction Dove Swarm Optimization-based Average Region Growing (CDSO-ARG). The segmented images are clustered into healthy and diseased categories by leveraging different color models. The important features are extracted from the diseased regions and subjected to the SS-PEDCNN classifier. Finally, the classifier identifies the diseases. Moreover, the detected diseases’ severity level is estimated utilizing POI with a fuzzy rule. The proposed technique’s efficacy is exemplified through experimental evaluations.</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

Rice plant disease detection using SS-PEDCNN with severity assessment

  • 期刊论文

  • 发表日期：

    2025年11月17日

查看全文
  • Preeti Yadav,
  • Parvinder Singh

摘要

Globally, rice is important as a primary food source. However, it is susceptible to diseases that cause significant production losses. For large farmlands, manual disease surveillance is impractical and costly. To resolve this, this paper proposes a novel framework called Scalar Softmax-based Positional Encoding Deep Convolutional Neural Network (SS-PEDCNN)-centric Rice Plant Disease (RPD) Detection framework with Severity Assessment capabilities by Percentage of Infections (POI) with Fuzzy Rule. For more efficient disease detection, the proposed system considers all parts of the rice plant, encompassing the stem, sheath, and leaf. Primarily, by utilizing the Correlated Log Transformed Double Plateau Histogram Equalization (CLT-DPHE), the contrast of input images is enhanced. Afterward, by employing the Shannon Entropy-based Gaussian Mixture Model (SE-GMM), backgrounds are removed. Then, the background removed images are segmented into separate parts (stem, sheath, and leaf) utilizing Constriction Dove Swarm Optimization-based Average Region Growing (CDSO-ARG). The segmented images are clustered into healthy and diseased categories by leveraging different color models. The important features are extracted from the diseased regions and subjected to the SS-PEDCNN classifier. Finally, the classifier identifies the diseases. Moreover, the detected diseases’ severity level is estimated utilizing POI with a fuzzy rule. The proposed technique’s efficacy is exemplified through experimental evaluations.