Meal Information Recognition Based on Smart Tableware Using Multiple Instance Learning

In recent years, people are paying more and more attention to the management of healthy meals because of the influence of lifestyle-related diseases, and some diet management systems are trying to help people lead a healthy life. Moreover, some studies have found that the proper meal habits can play a role in preventing disease to some extent. This paper introduces the smart tableware consisting of an acceleration sensor and a pressure sensor, which is used to obtain information about the sequence and content of meals. In addition, a method of analyzing and processing meal information through Multiple Instance Learning (MIL) is proposed to help people prevent diseases that are affected by lifestyle habits. At the same time, the acquisition process of MIL dataset is introduced and Support Vector Machine (SVM) is used. The performance evaluation results show that good results can be achieved by using MIL.