Scholarlink

<p>In every country, rice is a fundamental food crop, and it is susceptible to many different diseases. Diseases in rice plants are unfavourable factors that severely lower crop yield and quality. Even though it can be imprecise and take a lot of time, experienced biologists and farmers regularly observe plants for disease. Rice plant diseases can now be detected using computer vision technology and deep learning (DL), reducing the need for farmers to protect their harvests. In this study, an optimized DL approach is designed for detecting the disease in rice leaf. The first stage of preprocessing consists of three levels: altering the size, applying median filtering (MF) to remove noise, and utilizing Super-Resolution Convolutional Neural Network (SRCNN) to enhance rice images from low resolution to high resolution. The unhealthy area in the image of the rice plant was then identified using a modified grab cut segmentation procedure. After segmenting the data, a hybrid feature extraction technique was used to extract the features, which were then given to an optimized Bidirectional long short-term memory (Bi-LSTM) approach for disease classification in rice leaves. The proposed approach performance is tested using metrics Like precision, accuracy, recall, and specificity, obtaining 96.54%, 98.2%, 96.28%, and 98.78%. The proposed model’s obtained values are higher than those of the current methods. Thus, the proposed optimized deep learning approach for effectively detecting rice leaf disease with higher accuracy.</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

Modified grab cut segmentation based optimized Bi-LSTM Schema for rice leaf disease detection and classification

  • 期刊论文

  • 发表日期：

    2025年9月10日

查看全文
  • Devchand J. Chaudhari,
  • K. Malathi

摘要

In every country, rice is a fundamental food crop, and it is susceptible to many different diseases. Diseases in rice plants are unfavourable factors that severely lower crop yield and quality. Even though it can be imprecise and take a lot of time, experienced biologists and farmers regularly observe plants for disease. Rice plant diseases can now be detected using computer vision technology and deep learning (DL), reducing the need for farmers to protect their harvests. In this study, an optimized DL approach is designed for detecting the disease in rice leaf. The first stage of preprocessing consists of three levels: altering the size, applying median filtering (MF) to remove noise, and utilizing Super-Resolution Convolutional Neural Network (SRCNN) to enhance rice images from low resolution to high resolution. The unhealthy area in the image of the rice plant was then identified using a modified grab cut segmentation procedure. After segmenting the data, a hybrid feature extraction technique was used to extract the features, which were then given to an optimized Bidirectional long short-term memory (Bi-LSTM) approach for disease classification in rice leaves. The proposed approach performance is tested using metrics Like precision, accuracy, recall, and specificity, obtaining 96.54%, 98.2%, 96.28%, and 98.78%. The proposed model’s obtained values are higher than those of the current methods. Thus, the proposed optimized deep learning approach for effectively detecting rice leaf disease with higher accuracy.