Detection and Prediction of Pests Based on Leaf Images Using Efficient Net V2

This study proposes a new method for classifying pest images using the EfficientNetV2-L convolutional neural network architecture. The proposed model has a high degree of accuracy in identifying agricultural pests, such as ants, bees, beetles, and others, by utilizing transfer learning on a carefully selected dataset. This study enables efficient detection and timely interventions to reduce crop losses by fine-tuning pre-trained weights. The dataset, gathered from Flickr, guarantees diversity and realism in pictures of pests. The research findings highlight how machine learning may help advance sustainable agricultural methods by lowering the need for chemical pesticides. Finally, this study discusses about the advantages and challenges, providing guidance for further study and advancement in the field of pest image categorization.