Smartphone application for localization of real world objects, events and behaviors

Value Proposition

Conventional techniques for mapping the location of real-world objects or events are successful at identifying the location of static objects at a gross level. Mobile computing devices, such as smartphones, typically include a global positioning system (GPS) receiver for navigation and for displaying a position of the device on the map. However, localization of relatively small or time varying events, such as the location of places where individuals smoke in public, yield unsatisfactory results, especially if the observer cannot get close enough to the target of interest. Accordingly, there is a need to provide improved systems and techniques for defining the geographic position of real-world events, in a user's vicinity by quick and efficient means.


Researchers at Duke have developed a mobile application for defining a geographic position of a real-world object, event, or behavior. This app has a direct utility in determining the location of all places where individuals smoke in outdoor public environments. It uses mobile sensing and crowdsourced localization that essentially allow users to point to objects, or behavior without having knowledge of the address. For example, when user spots a smoking person, she or he may swipe on the smartphone screen in the direction of the event. Using this application, large number of observers could localize instances of smoking behavior and from this data, smoking probability map can be obtained to alert individuals of areas they might want to avoid. The application was built and tested on an Android smartphone platform.

Additional applications

The app has many potential applications for participatory reporting of events such as problems with infrastructure. Additionally, it can be used by institutions to evaluate effectiveness of smoking bans and restrictions.


  • A low-cost and near-continuous solution for event, behavior, and object reporting and localization
  • Does not require one to directly face the target and can be done from a vehicle
  • The app outperforms established approaches in terms of localization accuracy and detection coverage