Scholarlink

<p>Pests and diseases are a&#xa0;major problem that severely threaten walnut productivity across all phenological stages. Delays in intervention exacerbate economic risk, increase control costs, and intensify synergistic interactions between pests and host pathogens, amplifying environmental and harvest losses. Automatic pest and disease detection is essential, enabling real-time detection and interventions that limit pathogen intensification, minimize harvest loss, prevent spread to adjacent plants, and improve the efficiency and sustainability of crop protection systems. To address this multifactorial problem, the proposed study methodically compares and evaluates independent custom-layer-designed InceptionV3 and ResNet50 models for accurate pest and disease detection in walnut using the image dataset of diverse destructive pests comprising 17&#xa0;classes and a&#xa0;pathogen disease dataset comprising three classes, developed from farmer’s fields in and around Kupwara district of Jammu and Kashmir, India. Advanced preprocessing steps, including high-resolution image selection and data augmentation, were employed to enhance feature fidelity, improve model generalization, mitigate overfitting, and ensure robust model performance under diverse environmental settings. The proposed model performance is rigorously evaluated using accuracy, precision, recall, F1-score, and error matrix, with comprehensive benchmarking against state-of-the-art methods. For walnut pest detection, the experimental results demonstrate that the InceptionV3 model achieves superior performance compared to existing approaches, attaining training accuracy of 99.87%, validation accuracy of 98.62%, testing accuracy of 98.05%, and precision, recall, and F1-score of 98.07%. Similarly, for walnut disease detection, among the proposed models, InceptionV3 again attained enhanced detection ability, achieving training of 99.77%, validation accuracy of 98.68%, testing accuracy of 98.25%, with strong precision, recall, and F1-score of 98.67%, highlighting its robustness and effectiveness across disease detection. The proposed walnut pest and disease detection model can be integrated and deployed in real-world agricultural applications through Artificial Intelligence of Things (AIoT) and mobile applications for integrated pest and disease management (IPDM) in smart agriculture. Also, this study significantly contributes to the advancement of integrated sustainable development goals (SDGs) by enabling early pest and disease detection that supports timely, precise, and resource-efficient agricultural interventions. It contributes to SGD‑1 (No Poverty) by preventing severe food insecurity and alarming starvation, safeguarding the primary income source of millions of small-scale and marginalized farmers, through sustainable crop production. SGD‑2 (Zero Hunger) aims to minimize crop losses, ensure food security, and promote sustainable farming practices. SGD‑3 (Good Health and Well-being) aims to reduce the need for toxic chemical pesticides to protect farmers’ and consumers’ health and well-being. SGD-12 (Responsible Consumption and Production) is supported by maximizing yields and preventing postharvest losses, particularly through enabling SGD Target 12.4. It advances SGD-13 (Climate Action) by facilitating proactive prevention, fostering targeted and localized intervention rather than blanket spraying of pesticides. SGD-15 (Life on Land) is supported by promoting sustainable farming and land management practices, preserving biodiversity, and minimizing environmental degradation caused by pesticides overuse.</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

Precision Prevention and Early Detection of Active Walnut Pests and Diseases Through Deep Learning Applications for Sustainable Pomology

  • 期刊论文

  • 发表日期：

    2026年4月2日

查看全文
  • Khalil Ahmed,
  • Mithilesh Kumar Dubey

摘要

Pests and diseases are a major problem that severely threaten walnut productivity across all phenological stages. Delays in intervention exacerbate economic risk, increase control costs, and intensify synergistic interactions between pests and host pathogens, amplifying environmental and harvest losses. Automatic pest and disease detection is essential, enabling real-time detection and interventions that limit pathogen intensification, minimize harvest loss, prevent spread to adjacent plants, and improve the efficiency and sustainability of crop protection systems. To address this multifactorial problem, the proposed study methodically compares and evaluates independent custom-layer-designed InceptionV3 and ResNet50 models for accurate pest and disease detection in walnut using the image dataset of diverse destructive pests comprising 17 classes and a pathogen disease dataset comprising three classes, developed from farmer’s fields in and around Kupwara district of Jammu and Kashmir, India. Advanced preprocessing steps, including high-resolution image selection and data augmentation, were employed to enhance feature fidelity, improve model generalization, mitigate overfitting, and ensure robust model performance under diverse environmental settings. The proposed model performance is rigorously evaluated using accuracy, precision, recall, F1-score, and error matrix, with comprehensive benchmarking against state-of-the-art methods. For walnut pest detection, the experimental results demonstrate that the InceptionV3 model achieves superior performance compared to existing approaches, attaining training accuracy of 99.87%, validation accuracy of 98.62%, testing accuracy of 98.05%, and precision, recall, and F1-score of 98.07%. Similarly, for walnut disease detection, among the proposed models, InceptionV3 again attained enhanced detection ability, achieving training of 99.77%, validation accuracy of 98.68%, testing accuracy of 98.25%, with strong precision, recall, and F1-score of 98.67%, highlighting its robustness and effectiveness across disease detection. The proposed walnut pest and disease detection model can be integrated and deployed in real-world agricultural applications through Artificial Intelligence of Things (AIoT) and mobile applications for integrated pest and disease management (IPDM) in smart agriculture. Also, this study significantly contributes to the advancement of integrated sustainable development goals (SDGs) by enabling early pest and disease detection that supports timely, precise, and resource-efficient agricultural interventions. It contributes to SGD‑1 (No Poverty) by preventing severe food insecurity and alarming starvation, safeguarding the primary income source of millions of small-scale and marginalized farmers, through sustainable crop production. SGD‑2 (Zero Hunger) aims to minimize crop losses, ensure food security, and promote sustainable farming practices. SGD‑3 (Good Health and Well-being) aims to reduce the need for toxic chemical pesticides to protect farmers’ and consumers’ health and well-being. SGD-12 (Responsible Consumption and Production) is supported by maximizing yields and preventing postharvest losses, particularly through enabling SGD Target 12.4. It advances SGD-13 (Climate Action) by facilitating proactive prevention, fostering targeted and localized intervention rather than blanket spraying of pesticides. SGD-15 (Life on Land) is supported by promoting sustainable farming and land management practices, preserving biodiversity, and minimizing environmental degradation caused by pesticides overuse.