Scholarlink

<p>Early and accurate crop production prediction is essential for tactics for agricultural commodities and for double income for farmers. Deep learning has been effective in several agricultural mapping fields in recent years, such as crop type, crop area estimation, and classification. IoT provides resource traceability and monitoring in smart farming settings, allowing producers to enhance procedures, identify the origin of the output, and reassure clients of its high standard. Hence, this paper aims to utilize the ensemble deep learning architecture with IoT technology for crop yield prediction with the improved hybrid heuristic algorithm. Initially, the crop images are collected with IoT devices. Then, the deep features of crops are extracted from the 3-Dimensional Convolutional Neural Network (3DCNN), and taken as the first set of features. Further, the features from the soil parameters and environmental data are extracted and taken as second and third set features respectively. The extracted three sets of features are concatenated together and further, the concatenated feature is optimally selected using the hybrid optimization algorithm of Statistical Position of Hybrid Leader and Pathfinder (SPHLP) to improve the prediction performance. The selected optimal crop features are then given to the developed Optimal Ensemble Learning (OEL) method for predicting crop yield. This OEL network is the incorporation of a Support Vector Machine (SVM), Artificial Neural Network (ANN), Logistic Regression, and AdaBoost and it obtains the individual prediction score from all classifiers and then the weighted average score is obtained by considering all the prediction scores from SVM, ANN, Logistic Regression and Adaboost. To enhance the prediction performance, certain parameters in the OEL such as kernel size in SVM, hidden neuron count in ANN, and number of estimators in Adaboostare optimized with the same optimization algorithm of SPHLP. The comparison is carried out between the proposed crop yield prediction and health monitoring model and existing methods for showcasing the improved performance of the suggested model.</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

An efficient IoT-based crop yield prediction framework using optimal ensemble learning and hybridized optimization model

  • 期刊论文

  • 发表日期：

    2025年1月9日

查看全文
  • K. Shankar,
  • M. Moorthi

摘要

Early and accurate crop production prediction is essential for tactics for agricultural commodities and for double income for farmers. Deep learning has been effective in several agricultural mapping fields in recent years, such as crop type, crop area estimation, and classification. IoT provides resource traceability and monitoring in smart farming settings, allowing producers to enhance procedures, identify the origin of the output, and reassure clients of its high standard. Hence, this paper aims to utilize the ensemble deep learning architecture with IoT technology for crop yield prediction with the improved hybrid heuristic algorithm. Initially, the crop images are collected with IoT devices. Then, the deep features of crops are extracted from the 3-Dimensional Convolutional Neural Network (3DCNN), and taken as the first set of features. Further, the features from the soil parameters and environmental data are extracted and taken as second and third set features respectively. The extracted three sets of features are concatenated together and further, the concatenated feature is optimally selected using the hybrid optimization algorithm of Statistical Position of Hybrid Leader and Pathfinder (SPHLP) to improve the prediction performance. The selected optimal crop features are then given to the developed Optimal Ensemble Learning (OEL) method for predicting crop yield. This OEL network is the incorporation of a Support Vector Machine (SVM), Artificial Neural Network (ANN), Logistic Regression, and AdaBoost and it obtains the individual prediction score from all classifiers and then the weighted average score is obtained by considering all the prediction scores from SVM, ANN, Logistic Regression and Adaboost. To enhance the prediction performance, certain parameters in the OEL such as kernel size in SVM, hidden neuron count in ANN, and number of estimators in Adaboostare optimized with the same optimization algorithm of SPHLP. The comparison is carried out between the proposed crop yield prediction and health monitoring model and existing methods for showcasing the improved performance of the suggested model.