Abstract
Unbalanced lifestyle such as poor eating habits is known as a risk factor for many chronic diseases including obesity. In this respect, daily life monitoring could contribute crucial information to lifestyle coaching and diet management. In this report, we developed a method for recognition of three basic human activities, namely eating, drinking and speaking. Our recognition method used bone-conducted sound only to discriminate the three activities. The developed recognition algorithm used hierarchical clustering of sound spectra to distinguish sounds produced when masticating food, drnking liquid and speaking. The classification score for the three activities was as large as 80 %.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
