A DEEP LEARNING-BASED SMART MOBILE APPLICATION FOR AUTOMATED CITRUS FRUIT QUALITY CLASSIFICATION

Abstract

This study aims to develop a digital image-based citrus fruit quality detection system using the Convolutional Neural Network (CNN) method with the MobilenetV2 architecture and implement it into a cross-platform mobile application. The dataset used is a combination of public datasets and local citrus fruit datasets with five quality classes, namely fresh, half-ripe, black spots, damaged, and rotten. The total data is 5001 data, with 80% as training data and 20% as testing data. The CNN model was trained for 10 epochs and evaluated using accuracy, precision, recall, and F1 score metrics. The test results show that the CNN model is able to classify citrus fruit quality with high and consistent performance, indicated by precision, recall, and F1 score values in the range of 0.96-0.97. The trained model is integrated into a Flutter-based mobile application through the Flask backend, enabling real-time citrus fruit quality detection through a smartphone camera. The results of the study prove that the integration of CNN and cross-platform mobile applications can be an effective and objective solution in automatically detecting citrus fruit quality.