I recently worked on a set of Python scripts that can generate information based on geotagged photographs. The scripts are designed to take a search term and download a sample set of geotagged photos from flickr, using the flickr API. Then these photos are parsed out into different clusters based on location, color, date taken and other meta-data. From these clustered sets a call to the Google Places API is made to attempt to identify landmarks based on the locations where many photos are taken.

So, for example, with the search term San Francisco, the scripts return KML files of popular geotagged photos around San Francisco, but they also identify and return landmarks or points of interest in San Francisco (such as the Golden Gate Bridge, Alcatraz, Pier 41, Golden Gate Park, etc).


You can find the GitHub repository here.