Intelligent Pest Management System for Attaining Standards of Precision Agriculture

Abstract: A vast amount of potential exists in the agricultural industry to meet the growing demand for food while also providing nutritious and healthful meals. Plant bug identification is a difficult chore for farmers since a substantial amount of their crops is destroyed and the quality of their products is impaired as a result of pest assault. Traditional insect identification relies heavily on the expertise of taxonomists, who must be able to identify insects based only on their physical traits. Classifying insects in Wang and Xie datasets, we employed artificial neural networks (ANN), support vector machines (SVMs), KNNs, and naive Bayes. The convolutional neural network (2DCNN) model was also used to do this. The Wang and Xie datasets each contained nine and twenty-four insect classes, respectively. An insect pest detection system is shown here that uses foreground extraction and contour identification to identify insects in the presence of a highly complex background. The technique works well in experiments using the Wang, Xie, and Deng datasets. The models' classification accuracy was enhanced through the use of cross-validation with nine folds. Ac-cording to the findings, the 2DCNN model accurately categorized 91.5 percent of the nine class insects and 90 percent of the twenty-four class in-sects. To better detect insects, researchers used the insect pest detection method on datasets from Wang et al., Xie, Deng, and IP102. The approach was applied to four different datasets in total. Classification accuracy and calculation time were shown to be significantly improved when compared to present methods, and insects in field crops may be identified with higher efficiency than was previously achievable. Agronomists rely on classifica-tion accuracy results to detect agricultural insects at an early stage and reduce the amount of time it takes to boost crop yield and quality while cut-ting expenses.