Social Sensor Network

Our goal is to automatically search the ever increasing number of Web 2.0 websites where users openly share their photos and create a multi-resolution spatio-temporal coverage layer; automatically improving over time as camera resolutions improve and GPS inside cameras becomes commonplace. When a sufficient number of images have been aligned, it is simple to generate new viewpoints from different positions, orientations and at various times of the year.

This image illustrates a synthesis of 100 aligned photos, shared on Flickr around the Matterhorn, with the intensity of the heat-map weighted to reflect social remarks made my observers of the various images. Such a tool could provide tourists and photographer alike an interesting insight into where good photos could be taken.		From the photo alignment process, the exact orientation of the camera when the photo was taken can be derived and the appropriate kml file can be generated from Google Earth visualisation.		We have created a visual search engine to explore the mutual content of aligned photos. In 'Eagleeye' you can select subsets of images that share geo-content or find other images taken within an image.