Hunger Games Search Based Deep Convolutional Neural Network for Crop Pest Identification and Classification with Transfer Learning

Abstract

Agriculture has a great potential for improving the need of food and provides newlinenutritious and healthy food. Crop pest and the disease are essential factors to newlinedetermine the plants quality and production. The presence of crop pests is having newlinenegative effects on quality and quantity of agriculture products. When the pests are not newlineidentified in a proper time, the security of food is degraded. newlinePest attack damages the considerable part of plants and thus crop quality is newlinediminished. Farmers find very difficult to identify crop pest due to this reason. Further, newlineearly pest identification is a major complicated process in agricultural industry. The newlineinsects are the major reason for reducing the quality of crops and minimize the crop newlineproductivity. Therefore, to monitor and to evaluate the losses because of insects is newlineessential to provide better crop quality and agricultural safety. One of the easy ways newlinefor controlling the pest infection is pesticides usage. However, the excess utilization of newlinepesticides is harmful. Pest s identification is done by digital image processing, and it is newlineapplied to the field of agriculture, and it has huge benefit particularly in protection of newlineplants and has better improvements in crop management. The pest identification by newlinenaked eye is a general method and it is a time consuming. newlineDeep learning model have made remarkable achievements in the field of image newlineprocessing. Hence, to identify crop pest a novel hybrid classification model is newlineintroduced which identify the pest automatically. The proposed model has three stages newlinelike pre-processing, augmentation and classification. Initially, the pre-processing is newlinecarried out by the filtering technique ACF (Adaptive Cascaded Filter) for removing newlinethe noise and the quality of the image. This process is essential since the results are newlinemainly based on the pre-processing stage. Then the ACF is cascaded with DMF newline(Decision-based Median Filtering) and GIF (Guided Image Filtering) models. The newlinehigh dimensionality features are extracted