Human Physical Status detection related to Danger Situations based on Smartwatch and Smartphone

The life of an individual is characterized by daily activities such as going to the office or to school, working, doing sport. Alongside them, unpleasant situations related to criminal events, such as to be robbed, can take place. In this cases, it is difficult to ask for help or to contact the authorities about the ongoing criminal situation. In this context, on the basis of the behavior related to the human physical status, it is possible to exploit specific data, in order to extract relevant information that characterizes danger situations. Based on such idea, a model centered on three main aspects related to the human body posture, level of stress and geolocation, is proposed. The experimentation is based on a smartphone and a smartwatch, as they support both the data extraction its local computation centered on machine learning techniques. The paper presents the proposed approach along with a first evaluation.