Tuesday, October 25, 2011

How to do a reverse image search

There are two easy options for doing a reverse image search:
  • Tineye
  • Google Reverse Image Search
I guess most will be familiar with the former - but maybe not the latter?  The post summarises:
  • what is a reverse image search
  • why do a reverse image search
  • how Tin Eye works
  • how Google Reverse Image Search works
  • a comparison of the results
  • image privacy issues
It also highlights the results I got when doing a test search - using the "popular choice" image from an exhibition which had been highlighted on this blog.

What is a reverse image search?

A reverse image search uses no words.  To do a reverse image search you first introduce an image to a specialised search engine and then it looks for more images like the one you showed to it.

It's usual to be able to either identify the baseline image via a file or a URL.

Why do a reverse image search?

The main reasons for using a reverse image search are:
  • to find out where an image came from.  Given that the search is very likely to return a number of other sources for the image it may be take a while to identify the original source.
  • to find higher resolution images of an image - search results will tell you 
  • to locate internet sites where an image appears - the results provide a URL address
  • to track down places where your images appear on the Internet in order to enforce your rights under copyright law.  This may highlight places where:
    • your copyright is being infringed
    • copies of your images are being sold
    • your images have been altered and represented to be those of another person
  • to identify the owner of an image so you can ask to use it (which prompts the question of whether somebody would be able to get in touch with you to do this?)
  • to identify who has already used a publicly available image so as to avoid its use if used too much already.  This is particularly relevant to those using images they do not own on blogs and websites.
This article on Scientific American indicates that the improvements in reverse image search will lead to an increase in reports of copyright infringement which in turn will lead to a change in the copyright laws.  
The extent to which emerging laws will continue to protect artist’s livelihoods remains to be seen.

I guess a number of people are familiar with TinEye which is probably the best known reverse image search engine.

You upload an image or plug in a URL and it will tell you all the other places on the Internet that the image can also be seen.  It does not search for words, metadata or watermarks.  The way it works is a bit like comparing fingerprints.  It's looking for similar characteristics - when it finds enough of them it identifies a match.
You can submit an image to TinEye to find out where it came from, how it is being used, if modified versions of the image exist, or to find higher resolution versions.
Which is very useful if it happens to be one of your images and shouldn't be appearing anywhere else without your permission!

What's absolutely amazing is the number of images it compares it too
Searched over 2.0345 billion images in 2.450 seconds
Tineye is very accurate.  On the image I tested (provided to me by a gallery) TineEye found both the blogposts I used it on straight away.  It also found a number of other sites but only nine in total.  All but two were legitimate sites for an image which had been provided to me by a gallery.  Some were however using it as a larger size than the gallery allows - which was interesting!

However its accuracy may also understate the number of places the image can be found - see below.

If you want to know more:
  • check out this tutorial 
  • watch the introductory video
  • check out the Tineye plugins for Firefox, Chrome and Safari - so you can use it when appropriate straight from your browser.
  • check out the Tineye FAQs (very informative and helpful - much better than Google's equivalent)
    Google Reverse Search Engine

    This was introduced - very quietly - this summer.  I made a note of it at the time - but I've not heard a lot about it since then - and there's not a lot on the Internet about it - hence this post!

    Google Reverse Image Search - four ways to find an image
    You can search for an image in a number of different ways:
    • drag an image to the search cell (I've tried dragging from one window to another and that works fine)
    • upload an image - very straightforward
    • copy and paste the URL for an image - I found that doesn't work so well for an image on a blog.  It didn't recognise the Blogger URLs for example.
    • download the Chrome extension and install it (very fast) - which then allows you to right click an image (on a website or blog) and offers a choice to search Google with that image to identify where else it appears.  That works fine and is very fast.
    You can read more about how it works and also watch a video
    How it works
    Google uses computer vision techniques to match your image to other images in the Google Images index and additional image collections. From those matches, we try to generate an accurate "best guess" text description of your image, as well as find other images that have the same content as your search image. Your search results page can show results for that text description as well as related images.

    A comparison of the results

    My own experience was that Google found a great many more sites for my test image than Tin Eye did.  

    This experience is confirmed in this article on the SEO Whistleblower blog - Google rolls out reverse image search: RIP TinEye - the results generated by Google were way in excess of those generated by TinEye.

    My own feeling is that TinEye generates the most accurate results for the top few examples it returns while Google has strength in depth.  It can find many more examples of images just like the one you're searching on - and it can do it fast.

    Interestingly in an era when search engine results are becoming increasingly contaminated by those trying to "fix" the results, the image search gives ones of the cleanest and most accurate set of results I've seen for a long time.  I'll certainly be using it as my search of preference in relation to art history for example or when I'm trying to track down an artist's website.

    Image privacy

    There are a number of articles now (such as this one) which indicate that maybe the images that we think are private are not so private after all.  This is despite the fact that people should be aware of whether or not their images are being indexed for search purposes - particularly in relation to social media.

    The one thing I'd caution people about is to remember that all images are becoming increasingly searchable and that includes all the personal and (we may think) private images of ourselves and our children, families and friends.

    Never has it been truer that any image uploaded to a private site may be seen in other places on the Internet in no time at all.

    Thus a search on a face will turn up ALL the places that you've used that pic on the Internet.  Think about that one the next time somebody takes a photograph of you or you use a photograph on a social media site!

    Maybe try doing a reverse image search on a photo of yourself that you've used on the web?

    UPDATE (2015): You can read more about this topic on my my new website page How to do a reverse image search on Art Business Info. for Artists


    1. Fabulous post. There have been so many times I wished I could do this. Thank you, THANK YOU!

    2. I was not even aware of this search option! thank you very much for the post.

    3. I agree, wonderfully useful post. Many thanks, as usual, Katherine.

    4. Thank you - this post enabled me, in a few minutes, to find the artist responsible for an arresting piece of work I'd come across, uncredited, on a tumblr site.

      Useful for me, good for the artist.

    5. thanks Katherine... I have been able to find a few usages of an image of mine that I am NOT happy with.... the 10,000 reposting in tumblr is OK but the saving the image to use a backgrounds is NOT!

    6. I had a feeling there were quite a few people who hadn't realised there were these facilities for tracking down their images

      As Liz indicates there are, unfortunately, quite a few people who regard images as a "free good".

      Can I suggest people pay great attention to their settings on Flickr. I now have very limited sharing of images on Flickr after I discovered oodles of mine being used to decorate a spam blog which had a fair few adverts. Flickr positively promotes people lifting images and using them in other settings with no accreditation whatsoever for the originator.

      I now use it as an archive facility for thousands of images and a place to share for a very few.

    7. Thanks for the post, appreciated.

    8. You can also check out http://camfindapp.com/ that does reverse image search

    9. Might want to check out this reverse image search tool: http://www.mypicguard.com
      easy scan for a batch of images, find all copies of your images online, manage, track, and take action using built in integration to DMCA take down site.


    COMMENTS HAVE BEEN CLOSED AGAIN because of too much spam.
    My blog posts are always posted to my Making A Mark Facebook Page and you can comment there if you wish.

    Note: only a member of this blog may post a comment.