Investigation of deep learning for whitefly identification in coconut tree leaves

Abstract

ProblemWhiteflies are one of several pests that may damage coconut tree plantations, which are crucial to agriculture but susceptible to disease and pests. The diagnostic and therapeutic efficiency of conventional medical opinion may be limited. A scarcity of information in the agricultural sector has resulted from the fast development of technology. Drone photography provides a fresh viewpoint, but new methods are needed for tracking infectious diseases. AimThis research seeks to create a Deep Learning-assisted Whitefly Detection Model (DL-WDM) that can detect whiteflies on drone photos and thereby diagnose problems such as root bleeding, blade pollution, and insect infection in coconut trees. The objective is to improve illness diagnostics and plantation health using deep learning and computer vision. MethodsDrones were used to take pictures of both healthy and damaged coconut palms for this study. For model training, these pictures were used as the data set. The photos were analyzed for abnormal areas, such as those showing symptoms of illness or whitefly infestations, using popular segmentation algorithms. Images of coconut tree tops were analyzed using the VGG-16 model designed for this task. This paved the way for further illness categorization. Disease classification and prediction were the focus of training for a custom-built Deep Convolutional Neural Network (DCNN). Deep learning model can detect problems such as root bleeding, blade pollution, and insect infestation using the segmented areas. ResultsDL-WDM to track whiteflies and identify diseases was encouraging. Major findings include Locating irregular features in aerial photographs. The VGG-16 model identified coconut tree tops accurately—accurate illness diagnosis, including root bleeding, polluted blades, and insect infestation.