Assisted journey recollections from photo streams: (demo paper)

Abstract

We extract GPS traces from photo streams and analyze them to reveal movement types. Speed and locale patterns hint about the kinds of activity as captured by the photos of the day. When properly categorized and visualized, the photos and their movement patterns help people in navigating the itineraries of their past, and in retreating images of possible highlights. Our method is tolerant of erroneous and missing positional information in the photos' metadata. External geospatial resources can be further combined and visualized with the itineraries and photos to assist people's recollection of the places they were visiting.