Isn't revealing your location the whole point of taking a picture from your hotel window and sharing it with the world on Twitter? I wouldn't be too surprised if I posted "I'm in Miami, and here's a view from my Hotel window" and someone was able to tell me the Hotel name and which side I was staying on. Infact, I'd be more surprised if someone couldn't!<p>The techniques used are not particularly ingenious here, especially as most of the photo's seem to depict some obvious landmarks to provide an easy point of reference, especially when given the extra information to focus on a specific location.
Using the reflections as additional information is very interesting. However, I’m not that surprised that views from windows can be relatively easily located. Andrew Sullivan has been running his View From Your Window contest for four years now and people have been guessing the locations of the photos with scary accuracy (as good as in the linked post).<p>Here is the archive: <a href="http://dish.andrewsullivan.com/vfyw-contest/" rel="nofollow">http://dish.andrewsullivan.com/vfyw-contest/</a><p>Still, it’s probably more than 15 minutes work even in the best case scenario (and much, much longer in the worst case), can probably not be automated in any meaningful way (but that would be an interesting project, huh!) and is consequently only of interest to dedicated attackers, not general surveillance.
"Finally, do not forget that a reflection could be your enemy."<p>On a tangent - I didn't realize that the google maps 3D rendering has gotten so good over the last couple of years. I wonder how hard it would be to scrape the point data and satellite textures to source a GTA map, I'd love to cruise through actual Vice City with the GTA V engine.<p>If you haven't checked it out yourselves - breath-taking WebGL view of Miami (Chrome, OSX):<p><a href="https://www.google.com/maps/place/Miami,+FL/@25.7747883,-80.1751686,528a,35y,261.17h,71.49t/data=!3m1!1e3!4m2!3m1!1s0x88d9b0a20ec8c111:0xff96f271ddad4f65" rel="nofollow">https://www.google.com/maps/place/Miami,+FL/@25.7747883,-80....</a>
This reminds me, can't find it right now, but there is a picture circulating on Twitter of a safari park warning sign:<p>"Please make sure you turn off location and be careful about sharing photos of our rhino on social media, you may give away its location to poachers"
This reminds me of the child pornography case a few years back, where detectives were able to identify the hotels where the abuse took place via imagery of the rooms.<p>Interestingly, rather than figuring it out themselves (as the search space was too huge), they crowdsourced it by releasing images with the child photoshopped out.
Point is: when you post a view or a location name that's when you <i>want</i> to tell the world where you are. I'm not inadvertently revealing my location I'm very much advertently revealing it.<p>I assume there are much more creepy techniques to tell where I am when I <i>don't</i> explicitly reveal it.
If it comes to stalking someone, why just a hotel picture? Any picture, and previous context (e.g., "Miami, I am here"), can be tracked down to the exact location.
Views out the window are also very useful for finding the exact location of real estate for sale or rent, when the ad only shows an approximate location.
So, 99.9% of the information was gathered based on the actual photo, and .1% of it confirmed with reflections. Impressive. Hmm, this picture of a landmark came from the building near the landmark. How did you get that with just reflections?!?
I'll be taking this material and creating a tutorial "101: How to stalk a celebrity and make them feel slightly uncomfortable"<p>However, in all seriousness, nice levels of deduction going on here and only slightly creepy tweets :)
With a set of photos of the same static scene, it's possible to figure out the camera position, angle, and focal length of every camera, sometimes to millimeter accuracy.
You just have to correlate the same features in each frame, then solve it.<p>3d motion trackers do this. (Blender.org has one. You can solve this same problem using it. Bring in public and street view photos of the area. It's a bit of manual work to put all the images in a movie then go frame by frame and tag features... But you could get much more accurate than the article.)<p>The power is in having access to large databases of photos of everywhere, and even better if all the photos were taken the same way, like google's street view... and large amounts of computing resources to index and correlate all of the features from all of the images automatically.<p>I would be surprised if there is not a project in some huge company or lab that can identify features in nearly any public outdoor shot and determine exactly where it was taken. (perhaps for intelligence or forensics)<p>I believe this paper introduced me to the idea in 2006:
<a href="http://research.microsoft.com/en-us/um/redmond/groups/ivm/phototours/PhotoTourism.pdf" rel="nofollow">http://research.microsoft.com/en-us/um/redmond/groups/ivm/ph...</a>
This is hardly "inadvertently" revealing your location: (1) they've told you where they are (extra from the actual Tweets, not the photos) and (2) these places are all near well-known landscapes.<p>I mean good job on taking the time to actually do it but I doubt this could be done for the vast majority of photos like this.
As a countermeasure, the article notes turning off lights and so forth. Seems the author hasn't heard of polarized lenses. Others note that it's not a real security concern, no, but using a FLD lens will result in better photos when shooting through glass! Use it!
This only seems useful if it is automated and extremely fast. If you are law enforcement there are faster ways to do this. Like just calling in the city the "bad" guy last checked in at.<p>And if you aren't, there are much easier ways to get info on the room someone is using, like social engineering. Which I'd actually one if the examples the author gave when he simply asked the victim in twitter what floor he/she was on.<p>I haven't seen criminal scenarios where the bad guy takes a photo if his view or checks in on four square, that doesn't get rounded up fairly quickly.
There may come a time when indoor pictures could reveal your location as well. If you were to create a database of lots and lots of architectural plans, you could conceivably use computer vision algorithms for determining dimensions of a room as a few more known bits of entropy for identifying an indoor location.
From the title of the article I expected this to discuss a computer vision related technique for enhancing the camera location using reflections.<p>Using techniques similar to those in photosythn, you can already determine camera location fairly accurately and easier than the manual work done by the article's author.
Possibly more interesting:<p>Research shows eye-reflections in photos could be used to identify criminals:<p><a href="http://www.engadget.com/2013/12/29/eye-reflections-catch-criminals/" rel="nofollow">http://www.engadget.com/2013/12/29/eye-reflections-catch-cri...</a>
You might reveal your location but the person spying on you would have to be a stalker and a total creep, who basically have no life and nothing interesting to do. That's more scary than taking pictures with reflections IMO
Reminds me of the ridiculous enhancement processes in Hollywood. Except this time... it's real.<p><a href="https://www.youtube.com/watch?v=LhF_56SxrGk" rel="nofollow">https://www.youtube.com/watch?v=LhF_56SxrGk</a>
I would think a person would be more concerned about other things (not location) being relieved in the reflection. Like the strippers you have in the room. Or the drugs.
Also, taking photos of meals you are eating can inadvertently reveal your dietary choices.<p>Look at that burger - you can see from the menu that this is a vegetarian restaurant - that means that the 'meat' isn't really meat at all but a bean and nut mixture.
My takeaway: if you want to throw creepers like this guy off your trail, harvest a hotel picture from the internet and post it to your Twitter account.