Convolutional Neural Network-Based Image Classification for Improved Coconut Disease Identification

Coconut production is a vital source of income for the Philippine economy; thus, early detection of diseases is critical to improving the production capacity of coconut farmers. This project explores the use of deep learning technologies and computer vision to develop a model that can identify caterpillars, leaflets, drying leaflets, flaccidity, and yellowing in coconut trees. Specifically, we designed a deep 2D-Convolutional Neural Network (CNN) that can predict disease and pest infestations with high accuracy. The CNN was trained using a supervised deep-learning algorithm, which resulted in the five categories being predicted with an accuracy rate of 99% for yellowing and 100% for caterpillars, leaflets, drying of leaflets, and flaccidity, respectively. The proposed model has the potential to be a reliable diagnostic tool for effectively detecting diseases and infestations in coconut, supporting disease management programs in agriculture. As such, it is recommended that further research be conducted to integrate this technology into existing disease management programs for coconut farming in the Philippines.