Citrus Brand Classification by CNN Considering Load and Sound

This paper describes an intelligent system that classifies citrus brand considering physical features. The system consists of new original equipment to measure the citrus physical features and the convolutional neural network (CNN) to classify the citrus into any brand. When a needle probes citrus flesh, the load, and the sound are measured. From measured signals, the computer generates an image consisting of the sound spectrogram and the color bar expressing the load change. The image is inputted to CNN in order to infer the brand of citrus. In the experiment, CNN is validated.