Tineye - Search for visually similar images on the web

8 replies
I just started experimenting with TinEye.com which describes itself as follows:

"TinEye is the first image search engine on the web to use image identification technology. Given an image to search for, TinEye tells you where and how that image appears all over the web--even if it has been modified.

Just as you are familiar with entering text in a regular search engine such as Google to find web pages that contain that text, TinEye lets you submit an image to find web pages that contain that image."

I'm curious as to whether or not anyone else has used this search engine and what their experience has been?"

I know there is software that will search for duplicate images on your computer, but this is the first one that I've seen that searches the web.

I'm also curious if this is the only one or if there are others like it.
#duplicate #images #search #similar #tineye #visually #web
  • Profile picture of the author Sean Hoffman
    A couple of my friends are actually working on something similar right now.

    Looks neat but it sucks that you have to sign-up to use it...

    Google will soon buy them out
    {{ DiscussionBoard.errors[349439].message }}
    • Profile picture of the author psresearch
      Originally Posted by SEOSuperGuide View Post

      A couple of my friends are actually working on something similar right now.

      Looks neat but it sucks that you have to sign-up to use it...

      Google will soon buy them out
      Interesting. Please let us know when it's publicly available.
      {{ DiscussionBoard.errors[349511].message }}
      • Profile picture of the author Zaizen
        Thanks for the link . this will help me get better images/pictures.
        {{ DiscussionBoard.errors[349583].message }}
  • Profile picture of the author AverageGuy
    there are 2 kinds of image searching:
    1. tag, file name. like google image search. it actually search image name. has no relation with image itself.
    2. image indexing based on image itself. well, technology is very limited. basically, uses image features, such as color, edge, shape, .... to index images. it has being researched for many years, no break through yet. for practical usage, it still has a long way to go.

    watch the demo, it is cool to people who do not know much image processing (I mean research, not use photo shop), probably, it is interesting to some venture capital, but, it is not very impressive to me. but, it is always good to try. And good luck try to get some investment in this economy situation.



    david
    {{ DiscussionBoard.errors[349638].message }}
    • Profile picture of the author psresearch
      Originally Posted by AverageGuy View Post

      there are 2 kinds of image searching:
      1. tag, file name. like google image search. it actually search image name. has no relation with image itself.
      2. image indexing based on image itself. well, technology is very limited. basically, uses image features, such as color, edge, shape, .... to index images. it has being researched for many years, no break through yet. for practical usage, it still has a long way to go.

      watch the demo, it is cool to people who do not know much image processing (I mean research, not use photo shop), probably, it is interesting to some venture capital, but, it is not very impressive to me. but, it is always good to try. And good luck try to get some investment in this economy situation.



      david
      Thanks, David

      That's really helpful to know. Especially since TinEye is claiming that their technology will help you search for where your images are being used - sounds like that's not really going to happen any time soon.

      Do you know if the research ruled out that this type of technology COULD be developed? I mean is there some sort of limitation that has been proven scientifically impossible to get around?
      {{ DiscussionBoard.errors[349678].message }}
      • Profile picture of the author AverageGuy
        if your original image is not changed, sure it should find it. but if someone change it a little bit, can they find it?

        david

        Originally Posted by markquinn View Post

        That's really helpful to know. Especially since TinEye is claiming that their technology will help you search for where your images are being used - sounds like that's not really going to happen any time soon.
        {{ DiscussionBoard.errors[349723].message }}
        • Profile picture of the author psresearch
          Originally Posted by AverageGuy View Post

          if your original image is not changed, sure it should find it. but if someone change it a little bit, can they find it?

          david
          Ah, OK. Thanks for that clarification.

          You've been a huge help.

          Thanks.
          {{ DiscussionBoard.errors[349775].message }}
  • Profile picture of the author AverageGuy
    technology is amazing, IMHO, nothing can be proved "impossible". today, we say image indexing is diffcult, it does not mean in the future we can not do it. it is always possible in the future, but for now, image indexing is difficult, simply because image is much complex than text, image includes too much information which it is difficult for current computer technology to classify it. as old saying, one image worth one thousand words, actually, one image may have information more than 1 million words easily.

    so, for general purpose, indexing images are technical difficult for now. if, in certain controlled situation, it can work in certain degree. just like the demo, image size are very similar, the background are very similar, these two guys are standing in the similar position, then you can index them. it is just like if you input a very long keyword, search engine sure can find the pages that include these long keywords. but, what if the image is distorted, can the algorithm still work? that's the question.


    david
    {{ DiscussionBoard.errors[349721].message }}

Trending Topics