Convolutional Neural Networks for Estimating the Ripening State of Fuji Apples Using Visible and Near-Infrared Spectroscopy

Abstract The quality of fresh apple fruits is a major concern for consumers and manufacturers. Classification these according to their ripening stage one the most decisive factors in determining quality. In this regard, aim work develop new method non-destructive classification state Fuji apples using hyperspectral information visible near-infrared (Vis/NIR) regions. Spectra 172 samples range from 450 1000 nm were studied, which selected four different stages. A convolutional neural network (CNN) model was proposed perform samples. compared with three alternative methods based on artificial networks (ANN), support vector machines (SVM), k -nearest neighbors (KNN). results revealed that CNN outperformed methods, achieving correct rate (CCR) 96.5%, an average 89.5%, 95.93%, 91.68% ANN, SVM, KNN, respectively. These will help development device fast accurate estimation apples.