Like most Internet users, I’m a big fan of Google Image Search and like to keep up with what’s happening in the advancements in image recognition and indexing. It’s a step forward in artificial search intelligence, and can be extremely useful. 

TinEye is a new comparative image search tool from a Canadian image search specialist company called Idée Inc. Unlike Google Image Search, TinEye is designed to find duplicates or similar images from your image query, not text query – in fact, the two are barely comparable.  

I gave TinEye a bash, using it to locate the source of an unidentified image I had lying around. It's a simple sketch I had saved from some website ages ago without noting its source. Within a flash, I had the source of the image and a few variations. Not bad. 

Where In The World is Jackson Pollock?

I decided to give TinEye a run for its money. But what image would really test the capabilities of TinEye’s 'pattern recognition algorithms' indexing? Definitely one of those abstract Jackson Pollock paintings.  

To my surprise and delight, TinEye responded with 11 variations of Pollock’s Blue Poles. A search for the painting on Google Image Search results yielded almost 18 000 instances, so while TinEye's indexing figures are no where near those of Google, they are more relevant.

TinEye looked at the image itself, whereas Google’s indexing is based on image meta, relevant on site text and other elements (which often makes it susceptible to irrelevant image results).  

Still in the beta stages, TinEye isn’t necessarily a new concept - the FaceDouble Facebook application for instance is pretty good at recognising faces and creating matches. TinEye, however, is not designed to recognise faces and has a lot of really practical uses, rather than simply being for entertainment: 

• Sourcing images

• Finding similar images and variations

• Image research, and comparison

• Tracking images (think ORM)

There is also a browser plugin and a booklet app available on the site, and registration is free. Take a look at a few examples of how TinEye Image Search can come in handy.