Reverse Image Search :Who are the Players

The other day my colleague asked me an interesting question .He has interest in photography and some times he takes photographs of strange things without knowing what it is.Some times he doesn't even know what it is called in English.He asked me if we can have a service where I can upload an image and search for similar images to know more about it.

I did a bit of googling to find out if we already have something and that brought me two interesting services , TinEye and SAPIR.

Let us look at both of them to find out what they have to offer.

TinEye
You can upload an image to TinEye to know where it came from.It can find an image if a modified version of the image exists.It can not find similar images .To test this I tried a new image and it was not able to find any image .But when I tried a google logo the result was quite satisfactory.


According to TinEye
TinEye creates a unique and compact digital signature or 'fingerprint' for it, then compares this fingerprint to every other image in our index to retrieve matches. TinEye can even find a partial fingerprint match.TinEye does not typically find similar images (i.e. a different image with the same subject matter); it finds exact matches including those that have been cropped, edited or resized.

Though TinEye does not provide "similar image search" ,it is an amazing service which can find how and where the modified images are in use.

Look at this video to know more about TinEye.



SAPIR (Search in Audio-Visual Content Using Peer-to-peer Information Retrieval)

In SAPIR you can upload an image and search for it.It returns similar images if it finds one .It currently indexes flickr for images and the videos are provided by BBC.To test the service I uploaded a red apple and it returned me some positive encouraging results .Its back end is powered by CoPhIR (Content-based Photo Image Retrieval) and MUFIN (Multi-Feature Indexing Network).

This certainly looks interesting.According to readwriteweb ,IBM is testing SAPIR in collaboration with the European Union consortium.

It certainly is a good starting point.At present it only indexes structured images (flickr) for similar image search .It will be interesting to see if they can index unstructured images and can do a "similar image search" .
Check this video to know more about SAPIR.