An Investigation of Automatic Food Logging for Health Applications Based on Video Camera Glasses

Dietary monitoring can provide valuable information for individuals, while aggregated information from diet logs is important for public health. Although several food logging systems have been presented in the scientific literature, a small number have employed smartglasses. In this paper, we introduce eyewear-based software architecture for tag extraction from video by using several machine learning models. We demonstrate our software architecture with a system designed to help smart eyewear users to perform food logging in an automatic, passive way. Also, we conduct an experiment to report the performance of our system in various conditions.