Face and Hands Detection and Tracking Applied to the Monitoring of Medication Intake

This paper presents detection and tracking methods for userpsilas body parts in video sequences. They are applied to detect human activities, in the case of this work, the detection of medication intake. We use a technique based on color and shape to detect the body parts and the medication bottles. Color is used for skin detection, and the shape is used to distinguish the faces from the hands and differentiate bottles of medicine. To track these objects, we use methods based on color histograms, Hu moments and edges. For the recognition of medication intake, we use a Petri network and event recognition. Our method have an accuracy of more than 97% and allow the detection of the medication intake in various scenarios.