Texture estimation system of snacks using neural network considering sound and load

Abstract

This paper aims at construction of a system which estimates texture of snacks. The authors have rebuilt an equipment from the ground up in order to examine various foods. The system consists of an original equipment and a simple neural network model. The equipment examines the food by compressing it and observing load and sound simultaneously. The input of the neural network model is parameters expressing characteristics of the load change and the sound data. The model outputs numerical value ranged [0,1] representing the level of the textures such as “crunchiness’’ and “crispness’’. In order to validate the usefulness of the neural network model, the experiment is carried out. Three kinds of snacks such as rice crackers, potato chips and cookies are employed. The model estimates the appropriate texture value of the snacks which are not used for training the neural network model.