Detecting Calamansi Diseases through their Leaves Using Convolutional Neural Network

Calamansi is one of the major crops grown in the Philippines. Citrus diseases have been an issue in crop production worldwide, and these diseases can spread and infect other crops nearby. This study aims to implement a Convolutional Neural Network to detect the three calamansi diseases: Citrus Huanglongbing, Citrus Canker, and Citrus Blackspot, along with the healthy calamansi leaf. This paper shows a portable device using Raspberry Pi 4 that can detect calamansi disease with specific diseases: Datasets gathered online with the said disease were trained using the Python programming language with the libraries of TensorFlow and Keras, a library for image processing with the Convolutional Neural Network ResNet-50 pre-trained model to detect the said diseases. ResNet-50 is a type of Convolutional Neural Network used in image classification because of its high performance. A total of twenty (20) samples, five (5) each classification, were used for testing, and a confusion matrix was used to calculate the system's accuracy, which is 90%.