One Billion Cameras -

What Happens When A Billion People Worldwide Become Equipped With The Tools Of Visual Communication?

In the past four years the number of cameras in people’s hands worldwide has increased by 600%. It will double again over the next five years. Total number of cameras of all kinds sold in 2000: 85 Million units. Projected 2008 sales: One Billion units. Pervasive, web-connected imaging and the explosion of user-generated visual content are transforming how we communicate with each other in our personal, work, and community lives.

What happens when a billion users of many cultures, and scores of engineers and entrepreneurs, focus on new ways to enrich customers’ lives through imaging? What happens when R&D is fueled by an order-of-magnitude increase in economic reward? And what happens to the industry’s business models and value formulas?

FutureImage - 6Sight

More on Geotagging Statistics

"Every picture has its place" is the conceptual base from which geotagging provides an index to imagery. When you approach Fickr to see the "where are" all those millions of images, you don't get much of a picture or a map. The Flickr map above should show the distribution of all ~7.8 million images as of today. It does not.

What this map of all geotagged images (7.8 million plus) displays are recent additions aggregated by a clustering solution that operates at all scales of the Yahoo map. You do not get any feel for where all that imagery is - Flickr's world-wide image distribution?



When you change the scale by zooming in on a region of interest, the visualization of the image clusters becomes more refined. In this case, Big Bend National Park there is a central cluster marked as "Lots".... Man, is this slow. It works but its slow... lots of spatial sorting going on.



Zooming in again.... centered on the "Lots" cluster you now find yet another cluster set.

Yep, I am starting to get a feel for what might be going on. The zooming and panning control are really not well organized IMHO - slider bars simply suck. Waiting, waiting... slow!
Doesn't really do much for me.. you need to change the scale to get intimate with the positional excitement.





I am now zoomed in as close as Flickr can go. The central image cluster I am most interested in can now be found and selected. Yes, I get images and in this case they are all from a single author.

Maybe a mash-up might help? Lets try something more GIS-like?

Back in mid-September a Beau Gunderson used ArcGIS and a world-wide grid to count Flickr images with-in each grid area. Each grid was then grouped as to the total count and shaded according to the aggregate count in each individual grid. This created a 'choropleth" map. Its a visual statistic of ~2.5 million Flickr images where the more yellow grids have the higher counts of Flickr images falling in them. (see http://www.beaugunderson.com/flickr/ for the KML down load. Lets hope Beau runs an up date +5 million images later?)

What happens when we zoom in on Big National Park? Wow, this is fast...

Hmmmm, yep I see the park boundary but the Park is bisected by two grids - one counting 700+ images and the other only 30+ or so. The grid "resolution" breaks down at this scale. The label on the choropleth map certainly shows the aggregate count and is really fast but it needs a finer grid to help sort out where the pictures of interest are. Also the colorization/class by gird should indicate grid count difference but it does not. It's also out-of-date. Bummer!

What I am trying to lead you to is that there are two methods to providing a visual statistic that IMHO will be used to help organize the millions and soon billions of geotagged images being shared. One is more precise, the Flickr cluster method, and the second is far faster but has both a currency lag as well as break-down of visual comparison that fails at a larger scale when its grid and its classification do not fit a finer resolution requirement.

"Vector is more corrector... but raster is faster" ... GIS insight from Dr. Joe Berry (http://www.innovativegis.com/basis/).

Geotagging Statistics

Was curious about the "total count" progress for geotagging imagery. Trying to find direct counts as well as proxies. I went through my notes and here are some quick observations.

Flickr started formally "geotagging" images on or about the 28th of August. They suggest they have around 300 million images; all in around 7.8 million are geo tagged as of the end of November 2006.

August 28th - the start - assume 0
August 29th - 1.234,284 million
September 9th - around 2.5 million
November 20th - 7.340,532 million
December 1st - 7.848,101 million

On average this works out to around 80,000 images per day for 93 days or so... seems to be declining a bit from one million per day at opening to a bit over 50k in recent days. Still a heap.

Also ran on to a stat from WebShots... they have around 397,000,000 images as of end of November. They have quite a good geographical organization in that there is a long list of images organized by goeraphical name - but no obvious map yet...

Kodak's Easyshare is rumored to be approaching a billion total images. Nothing obvious as to location though but I have not spent much time digging into what they are doing.

Zooomr is also hunting the geotagger. The have an interestig map up but just now it was a bit jammed..

Any other insights to the geotagging of imagery?