An analysis of eating activities for automatic food type recognition

Nowadays, chronic diseases such as type 2 diabetes or cardiovascular diseases are considered to be one of the most serious threats to healthy life. These kinds of diseases are primarily caused by an unhealthy lifestyle including lack of exercise, irregular meal patterns and abuse of addictive substances such as alcohol, caffeine and nicotine. Therefore, observing our daily lives is crucial in developing interventions to reduce the risk of lifestyle diseases. In order to manage and predict progression of diseases of a patient, objective measurement of lifestyle is essential. However, self-reporting questionnaires and interviews have limitation due to human errors and difficulty of conducting. In this paper, we analysed users´ eating activities and comprising sub-actions for developing eating activity recognition system based on a tri-axial accelerometer embedded wrist band. By analysing actions in eating activities, we can improve the accuracy of the recognition of eating activities and also provide clues that identifying the type of foods.