Metadata-based photo filtering for facility management

Facility management information systems are difficult to manage due to the enormous scope of buildings. We present a method for organizing and retrieving photos from massive facility management photo databases using photo-metadata. Providing an intuitive organizational scheme, location, camera orientation, and semantic content description metadata exploit the physical structure of the existing building their photos represent. Location is inferred using Bluetooth beacons. Camera orientation is inferred from the device's on-board inertial measurement unit (IMU). Image content is inferred using automatic object recognition. As a user traverses the interior of a facility, the photo library displayed on their mobile device is ordered by the proximity to the user's current location. A root mean square error (RMSE) of 1.1 meters is achieved for localization. This ordering is refined using a query image's camera orientation and image semantic content description. The presented photo management application reduces the time required to find relevant historical photos representing objects within a user's vicinity.