Development of Web-based Chili Plant Diseases Identification System

Abstract

Chili is one of the high economic vegetables. However, chili diseases are considered one of the main factors influencing its production. To minimize losses in production, it is essential to develop a tool to assist the farmers identify the disease. Therefore, in this work, the development of web-based chili plant disease identification system is developed. This system uses Convolutional Neural Network (CNN) model to identify healthy and diseased (Bacterial Spot and Whiteflies) chili based on its leaf image. All chili image datasets were obtained from Kaggle and it contains 1000 images per class. Besides that, this study also investigates the accuracy performance of the model based on the number of images dataset (250 and 1000 per class) for epochs of 50. The findings show that the CNN model with 50 epochs and 1000 images per class gives better accuracy performance on the system.  By using these findings, a local host server is developed using the ReactJS user interface with a CNN-based identification system as a core. The aim of the proposed solution is able to assist the farmers to monitor the health of their plants.