Well in a latest paper presented by Google stated that they are in a idea to conduct a research project that will help their bots to recognize images of 50,000 landmarks all around the world with 80% accuracy. Google also says that this is just a research project but not a new service.
A computer cannot recognize a picture of any landmark like a human but its very nice to hear about the technology that can do that as well. Google says it will be a great boon for travelers if they can actually implement and boost the accuracy rate of identifying the picture of a landmark.
What is this all about?
If you have a camera with Internet enabled, you can just take a snap of the landmark; ask google if they can find any info about it. That would be really cool if google can fetch you some info.
How does this work?
Google filtered huge number of images from Picasa and Panoramio, and “pruned” using efficient image matching and unsupervised clustering techniques. The final step includes indexing system for fast image recognition which can compare the image you upload to the standard images in google image index. Once a match is found google displays the results. How cool is that if this really works.
<Source>
Subscribe Updates, Its FREE!
→ 
→ 
{ 3 comments… read them below or add one }
Google rocks
what could be the potential use of this technology?
Well just for the begining, when we search for some images on google then we type the key text.
Example:
searching for: wallpapers which promote eco frendiliness, or some thing green, or informative.
Possible key words: eco wall papers, green wallpapers
Results in google images: wallpapers which have leaves and decoratives or wallpapers which are pretty much green in colour.
Think if we can tell the computer what we want to see rather than a trial and error methodology.