Automated Pest Classification of Paddy Field using Drone Imagery

Pests are a real problem to agriculture especially the paddy fields in that they contribute greatly to losses in agricultural productivity. The present research presents an innovative approach for unearthing the topology of pest-infected rice crops through a classification of a periodic series of images captured on a drone. The system consists of image processing procedures and deep learning algorithms to scan images of high spatial resolution obtained from drones. The method integration plan comprises pretreatment of images acquired from the drone, feature extraction by CNNs, and pest classification by a multi-class classification model. It is adept at distinguishing between general pest species that infest paddy fields to allow early, targeted pest control. They also bring the added value of scalability and accessibility, which make it possible to implement large scale pest monitoring. Performance evaluations prove that the system is very resistant and precise thus showing its ability to change the current pest control strategies used in farming. This minimizes the reliance on the inspection method and at the same time supports sustainable agriculture through development of constructive pest control strategies.