CLASSIFICATION OF MANGO FRUIT MATURITY USING SUPPORT VECTOR MACHINE (SVM) AND CONVOLUTION NEURAL NETWORK (CNN)

Abstract

Mango ripeness is one of the main indicators of fruit quality, which is generally still assessed subjectively through visual observation. Therefore, an accurate and objective automatic classification system is needed. This study aims to classify the ripeness level of mangoes based on digital images using a Convolutional Neural Network (CNN) and a combination of CNN–Support Vector Machine (SVM). The dataset used consists of 198 colored (RGB) mango images divided into three classes, namely unripe mangoes, ripe mangoes, and rotten mangoes. The pre-processing stage includes changing the image size to 128×128 pixels and normalizing the pixel values. CNN is used to extract visual features as well as an initial classifier. Next, the features extracted by CNN were used as input to the SVM model with a Radial Basis Function (RBF) kernel to improve classification performance. Evaluation was performed using accuracy, confusion matrix, and precision, recall, and f1-score values. The test results showed that the CNN model achieved a training accuracy of 96.38% and a validation accuracy of 97.47%. The CNN–SVM approach provided better results with a classification accuracy of 99% and a very low error rate across all classes. Based on these results, it can be concluded that the combination of CNN and SVM is effective for classifying the ripeness of mangoes and has the potential to be applied to automatic fruit sorting systems.