Key points are not available for this paper at this time.
We introduce MunchSonic, an AI-powered active acoustic sensing system integrated into eyeglasses, designed to track fine-grained dietary actions like hand-to-mouth movements for food intake, chewing, and drinking. MunchSonic emits inaudible ultrasonic waves from a commodity eyeglass frame. The reflected signals contain rich information about the position and movements of various body parts, including the mouth, jaw, arms, and hands, all of which are involved in eating activities. These signals are then processed by a custom deep-learning pipeline to classify six actions: food intake, chewing, drinking, talking, face-hand touching, and other activities (null). In an unconstrained user study with 12 participants, MunchSonic achieves a 93.5% macro F1-score in a user-independent evaluation with a 2-second time resolution, demonstrating its effectiveness. Additionally, MunchSonic accurately tracks eating episodes and the frequency of food intake within those episodes.
Mahmud et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: