Mapping the World’s Photos
Posted: April 27th, 2009 | 1 Comment »Related to my past project on Tracing the Visitor’s Eye and World’s Eyes, David Crandall and colleagues at Cornell have created global and city maps and identify popular snapping sites from photos uploaded on Flickr. They ran statistical analyses to identify the most important clusters of “”what the world is paying attention to”. Next they analysed the text tags added to photographs in those clusters, as well as key visual features from each image, to automatically find the world’s most interesting tourist sites. We find that visual and temporal features improve the ability to estimate the location of a photo, compared to using just textual features (slightly similar to IM2GPS: estimating geographic information from a single image). Their paper, Mapping the World’s Photos, recently presented at the WWW 2009 conference explains the process and results in more details.
Representative images for the top landmark in each of the top 20 European cities. All parts of the figure, including the representative images, textual labels, and even the map itself were produced automatically from our corpus of geo-tagged photos. Image courtesy of David Crandall. More on the project’s web page.
Relation to my thesis: Similar to my work, Crandall et al. I have used the spatial distribution of where people take photos to define a relation between the photos that are taken at popular places. I have used similar representations of spatio-temporal traces to illustrate my techniques and finding. However, my motivations has not been to improve photo management and organization applications (see also Mor Namaan’s work). Similar to Currid and Williams’ work on Mapping the Cultural Buzz, I employed georeferenced photos as a proxy to “qualify” the space (e.g. its attractiveness). But rather than mapping and qualifying a timeless space (Currid and Williams talk about “buzz-worthy” without referring to when), I particularly explore the temporal aspect of the data (see The End of Timelessness). For instance, in my work on the NYC Waterfalls, I studies the evolution of the density of georeferenced photos and the flows of photographers to provide evidences of the evolution of the attractiveness of the space (journal paper still in submission).
The Visualization of photographer movement in Manhattan mapped by Crandall is particularly stiking:
To produce these figures, we plotted the geolocated coordinates of sequences of images taken by the same user, sorted by time, for which consecutive photos were no more than 30 minutes apart. Image courtesy of David Crandall
I have just posted one entry at my blog. It is about the map of Gripe Porcina. It proves that Spain is the entrance of Europe for Mexico.
When ex KGB guy was assassinated in London by radioactivity, the most contaminated area was Barcelona. It is because London’s airplain is most connected with BCN, thanks to the cheap economy, Click Air, Ryanair, etc.
The essential question of this time is the same. Mexico is most connected with BCN, because of the culture, langueage, etc….and now, we can know it through new technologies at real time.
Now, we have new tool, thanks to the emerging technology, and the question is how we can use it. I hope that we can solve this actual crisis by our new tool!