Scholarlink

<p>Plant disease diagnosis in the agricultural field is a main concern that includes factors impacting crop productivity, deficiency and different diseases. Early and accurate identifications are taken as primary challenges in plant disease detection, serving as precautionary measures for preventing the crop field from attacks. The significant challenges addressed in this research work are extraction of a limited number of features, manual identification, complexity issues, lack of a large set of data and lack of region identification. To eliminate these challenges during the detection and classification of rice plant diseases, an efficient Artificial Layered Fuzzy Neural Network based African Vulture Optimization algorithm is proposed. Here, different types of images are gathered from various plant leaf disease identification related data sources that are preprocessed to improve image quality, reduce noise, and enhance contrast while addressing overfitting. The preprocessed images are included in the feature extraction model named as Cross Fusionformer for extracting more informative features presented in the rice plant images. The extracted features from different modalities are fused through the Squeeze Excitation model which fuses the weights of the features for enhancing network performances and minimizing feature redundancy. The affected and healthy regions of the plants are segmented according to the extracted features by using Spatial Fuzzy C-Means process. During the accurate identification and classification of rice plant diseases, the Artificial Layered Depth Separable Neural Network model is implemented. The Modified Depthwise Separable Convolutional layer is additionally included in the detection model as a residual connection for minimizing a number of parameters as well as complexity while enhancing generalization ability of the model. The classification model displayed the multiclass classification outcome from the collected different rice plant images. The Differential Bitwise African Vultures Optimization Algorithm tunes the hyperparameters presented in the proposed model by eliminating local optimum issues and enhancing the searching ability. The comprehensive analysis predicted the outcomes of highest accuracy of 98.87% and the lowest execution time of 0.09&#xa0;min from the proposed model. This research article achieves the superior performance of the proposed model is estimated through involving various classification measures when compared to existing analyses.</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 comprehensive cross-attention and fuzzy segmentation approach for rice plant disease detection

  • 期刊论文

  • 发表日期：

    2025年8月28日

查看全文
  • R. Prabavathi,
  • Balika J. Chelliah,
  • R. Sampath,
  • S. Vinodh Kumar

摘要

Plant disease diagnosis in the agricultural field is a main concern that includes factors impacting crop productivity, deficiency and different diseases. Early and accurate identifications are taken as primary challenges in plant disease detection, serving as precautionary measures for preventing the crop field from attacks. The significant challenges addressed in this research work are extraction of a limited number of features, manual identification, complexity issues, lack of a large set of data and lack of region identification. To eliminate these challenges during the detection and classification of rice plant diseases, an efficient Artificial Layered Fuzzy Neural Network based African Vulture Optimization algorithm is proposed. Here, different types of images are gathered from various plant leaf disease identification related data sources that are preprocessed to improve image quality, reduce noise, and enhance contrast while addressing overfitting. The preprocessed images are included in the feature extraction model named as Cross Fusionformer for extracting more informative features presented in the rice plant images. The extracted features from different modalities are fused through the Squeeze Excitation model which fuses the weights of the features for enhancing network performances and minimizing feature redundancy. The affected and healthy regions of the plants are segmented according to the extracted features by using Spatial Fuzzy C-Means process. During the accurate identification and classification of rice plant diseases, the Artificial Layered Depth Separable Neural Network model is implemented. The Modified Depthwise Separable Convolutional layer is additionally included in the detection model as a residual connection for minimizing a number of parameters as well as complexity while enhancing generalization ability of the model. The classification model displayed the multiclass classification outcome from the collected different rice plant images. The Differential Bitwise African Vultures Optimization Algorithm tunes the hyperparameters presented in the proposed model by eliminating local optimum issues and enhancing the searching ability. The comprehensive analysis predicted the outcomes of highest accuracy of 98.87% and the lowest execution time of 0.09 min from the proposed model. This research article achieves the superior performance of the proposed model is estimated through involving various classification measures when compared to existing analyses.