I was recently surprised to learn that you now could search by images in Google.
Google users will be able to search by talking to a computer or dropping an image in the search box. Users can paste an image’s web address, upload a photo from their computers, drag and drop an image into the search box or use Chrome or Firefox extensions. Apparently, Google will analyzes the fundamental features of the image (lines and shapes), and match it with results in its database of images. It will only use face recognition when the picture is of a celebrity and already exists on the public Web.
I am not sure if it will be able to recognize the images of my university apartment but anyway, I think it’s pretty interesting. And it makes it even harder for competitors like Bing and Yahoo to make up any of the distant ground between them and Google.
I remember Professor Kagan saying in class something like images would not generate good traffic unless you used the proper keywords in the tags (if I recall properly). I now wonder if these new features will affect the search optimization process.
See also:
2 comments:
I wanted to make some further comments to enforce the idea that image search will further strengthen the position of Google in the search engine panorama.
The cited technology is called reverse image search engine. There are a number of websites today offering the service of taking an image as input and providing all similar images. If the image is well known already (e.g. painting, celebrity) these engines work particularly well, otherwise they try to use their algorithms based on line and color proximity, but today results are still fairly poor.
Anyway the point that I am making is: if I have to look for an image that I already have and find the same image on the web with a higher resolution I would use TinEye a quite powerful tool for this kind of search. (for other reverse image search engine providers see: http://www.comptalks.com/top-10-reverse-image-search-engines/#comments).
On the other hand more often I am interested in finding content related to an image (e.g. I take a photo of a paiting in an art gallery and would like to find out more etc.). In this case the engine that can give me the most added value is again google: combining image search with content! As a matter of fact the real added value for me is not that of finding a number of identical capies of the image that I uploaded but to understand who the painter is and what is the story of the painting...
In conclusion I believe that reverse image search on its own has limited potential: the integration with the search platform will become a must for most companies. So just one year from the creation of a number of companies which were intending to revolutionize the panorama of search engines, google is again taking over by integrating this idea into its traditional search engine. How hard it is to compete with the giant!
I completely agree @Andrea. I think the potential of finding content related to images is HUGE and just beginning to be explored. One interesting example of this that jumps to mind: I was recently talking a fellow entrepreneur who is in the process of creating an app which is like a "sommelier in your pocket" -- in other words, this app (mainly meant for people who are not wine connoisseurs but who have an interest in learning about wine generally and specifically which wines they like) allows people to input every wine they like, dislike, etc. over a period of time. Gradually, the app then learns which kinds of wines a person likes, and eventually is able to suggest wines for that person. How does reverse image search come into play? Well, this entrepreneur happens to work at Google, and so knows about the latest developments with Google's image search index. So what he is planning on doing is adding a feature to his app where a person can take a photo of the label on a bottle of wine (at a restaurant or wine shop) and the app (using Google's reverse image technology) is able to figure out what kind of wine that is, and make a recommendation about it accordingly. How cool is that?! In my view, this wine app is just one example of reverse image search combined with a related content search. It's crazy to imagine what the mobile apps will start to look like as this technology improves (and I'm sure there will be plenty of privacy issues once you consider reverse image searches of people's faces, etc.). But for now I'm mainly excited about this potential.
Post a Comment