Recognition of meal information using recurrent neural network and gated recurrent unit

With the development of the Internet of Things technology, smart devices and systems for collecting and detecting meal information are becoming popular. In this research, multiple sensors are attached to the tableware to detect meal information. Since the data obtained from the smart tableware is sequential, algorithms such as recurrent neural network are used to detect meal information. In addition, multi-instance learning is also used to recognize meal information. The accuracy of different algorithms is compared and analysed in the experiment, and multi-instance learning achieves higher accuracy in the recognition of meal information. Furthermore, meal information can be browsed, and users can browse processed meal information and suggestions through the system. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

This research involves attaching multiple sensors to tableware to detect meal information, using algorithms such as recurrent neural network and multi-instance learning to achieve high accuracy. Users can browse processed meal information and suggestions through the system.