Pomelo Orange Desease Detection Using CNN Based on Digital Image Processing

Abstract

Pomelo is one of the important fruits in the agricultural industry and has high commercial value. However, pomelos are susceptible to disease. Detecting diseases in pomelos is crucial for maintaining the quality and quantity of production. However, disease symptoms in pomelos are often complex and difficult to accurately detect through human visual observation. Therefore, image processing is a solution for detecting diseases in pomelos. Convolutional Neural Network (CNN) is a type of artificial neural network architecture that is highly effective in analyzing and predicting diseases. In this study, a model is designed and built to detect and classify diseases in pomelos based on their skin. The study uses a dataset obtained from self-documentation using a digital camera, which includes images of Diplodia/Blondok, Cancer, Fruit Fly, and healthy pomelos. In the preprocessing stage, the dataset is divided into training, validation, and testing data. Feature extraction is also performed using thresholding, contour detection, and bounding boxes. During the model processing stage, a model is created using training data, validation data, hyperparameters, and transfer learning. The results of the study show that the DensNet121 model achieved an accuracy of 92%.