Automatic Detection of Chewing and Swallowing Using Attention-Based Fusion
Akihiro Nakamura, Takato Saito, Daizo Ikeda, K. Ohta, Hiroshi Mineno, Masafumi Nishimura · 2021 IEEE 10th Global Conference on Consumer Electronics (GCCE) · 2021
A series of eating behaviors, including chewing and swallowing, is deemed critical for the maintenance of good health. Therefore, a system that automatically detects eating behaviors must be constructed, using easily collected sound information. This study developed a new attention-based fusion method to integrate data compiled from multiple sound sources and obtained effective results. In addition, it achieved performance enhancements in comparison to simply early and late fusion methods, and the developed technique could more accurately capture eating behaviors. Sufficient detection performance is expected to be attained for the development of an automatic monitoring system for eating behaviors.