Spectrogram-based audio classification of nutrition intake

Abstract

Acoustic monitoring of food intake in an unobtrusive, wearable form-factor can encourage healthy dietary choices by enabling individuals to monitor their eating patterns, maintain regularity in their meal times, and ensure adequate hydration levels. In this paper, we describe a system capable of monitoring food intake by means of a throat microphone, classifying the data based on the food being consumed among several categories through spectrogram analysis, and providing user feedback in the form of mobile application. We are able to classify sandwich swallows, sandwich chewing, water swallows, and none, with an F-Measure of 0.836.