JIITA, vol. 1, no. 1, pp.1-5, Jun, 2017
Automatic Food Intake Frequency Detection Method
Jinwoo Jo, Ahyoung Choi
Department of Software, Gachon University, Seongnam, South Korea
Abstract: Recently many products or services are being introduced to monitor calories through analyzing user’s food intake behavior. However, it provided low usability that users should write dietary information manually and some cannot be used in a normal daily life. In this work, we proposed automatic food intake frequency detection method based on accelerometer and gyroscope sensors of smart watches. This system detected wrist motion of eating activity and recognized food intake frequency in real time. For this particular application, we selected two dominant features such as an average sum of accelerometer data and a yaw value of gyroscope data and developed algorithm to count eating behavior. We evaluated this method in a pilot study and we found that accuracy of this method was about 90 % with 20 subjects’ data in an experimental setting.
Keyword: activity recognition; accelerometer and gyroscope sensors; obesity management