An Efficient Approach for Crops Pests Recognition and Classification Based on Novel DeepPestNet Deep Learning Model

Crop pests are to blame for significant economic, social, and environmental losses worldwide. Various have different control strategies, precisely identifying has become crucial pest is a difficulty in agriculture. Many agricultural professionals interested deep learning (DL) models since they shown promise image recognition. Pest identification approaches literature relatively low accuracy recognition classification due the complexity of their algorithms limited data availability. Misclassification insect sometimes leads using wrong pesticides, causing harm yields surrounding environment. It necessitates developing an automated system capable more accurate classification. This paper presents novel end-to-end DeepPestNet framework The proposed model 11 learnable layers, including eight convolutional three fully connected (FC) layers. We used rotations techniques increase size dataset augmentations test generalizability approach. popular Deng’s crops set assess framework. method recognize classify crop into 10-class pests, i.e., Locusta migratoria, Euproctis pseudoconspersa strand, chrysochus Chinensis, empoasca flavescens, Spodoptera exigua, larva laspeyresia pomonella, parasa lepida, acrida cinerea, S. L.pomonella types insects pests. achieved optimal 100%. compared approach with traditional pre-trained models. To verify general adaptability this model, we tested on standard Kaggle “Pest Dataset” nine pests: aphids, armyworm, beetle, bollworm, grasshopper, mites, mosquito, sawfly, stem borer 98.92%. can provide specialists farmers immediate effective aid recognizing potentially reducing economic yield losses.