Maturity Status Estimation of Banana Using Image Deep Feature and Parallel Feature Fusion

Abstract

AbstractImmediately following mango, the banana (Musa sp.) is India’s second most important fruit crop. Once again, bananas are the most important fruit in worldwide trade and the most widely consumed, ranking second only to citrus in terms of value after apples. The size, colour, and ripeness of the fruits are the primary factors in grading. The grading of bananas is based on maturity in four stages: green, yellowish-green, mid-ripen, and overripen. Here, the maturity label of banana is estimated using deep features of VGG16 and texture feature with SVM classifier. The performance of classification is measured individually for deep feature and texture features. The classification task is also performed using both deep and texture features using parallel fusion. The accuracy and AUC using deep feature, texture feature and both (using parallel feature fusion) are 92.34% and 0.99%, 89.99% and 0.97%, and 99.87% and 1%, respectively.KeywordsBanana gradingDeep featureParallel feature fusionVGG16Texture feature