Automatic identification of drinking activities at home using depth data from RGB-D camera

With the increasing elderly population all around the world, the healthcare services to elderly has become more important. Dementia is one of the common cognitive problems among the elderly population where a basic daily dining activity could be difficult for them to perform. Drinking activities is one of the most important daily needs to keep a person away from dehydrating. Most of the existing works on dining activities recognitions are mainly based on wearable sensors and sensor rich eating utensil. Although the accuracy for sensor based techniques is high, some of the elderly might be reluctant to wear, and some might even forgot to wear altogether. In this paper, we propose a novel system that is purely based on depth data from an RGB-D camera for drinking activities recognition. Dynamic time warping algorithm is used to recognize and detect the drinking activities.