Image Classification Model for Data Sentry of Plant Diseases Using Convolutional Neural Networks

This study examines the effectiveness of using machine learning-based image recognition model for classifying common diseases in crops. This study addresses the critical need for swift and accurate disease identification in agriculture, particularly focusing on enhancing food security by preserving crop yield through early intervention. The study employs a dataset from the study of Mafi, et. al., about eggplant disease recognition composed of thousands of images representing seven distinct eggplant disease classes, with each class containing approximately 1600 images [1]. Through extensive pre-processing and image augmentation techniques, the dataset is standardized to ensure compatibility with the model. Additionally, a data acquisition framework was developed, integrating frontend capabilities to highlight image capturing and database as sensor data storage. A graphical user interface (GUI) is also designed to provide stakeholders with practical tools for swift disease identification. The convolutional neural network (CNN) model achieves an average accuracy of 94 % in classifying diseases, indicating its effectiveness in recognizing plant disease. Overall, this research contributes to the advancement of agricultural productivity and the reduction of crop losses through the capabilities of a machine learning method.