- Spot the iPhone!
Over a year ago Jason Kottke photoshopped a Windows desktop onto an iPhone and posted it on his very popular blog. Yesterday it was spotted on an Australian TV news show. Today, I used TinEye to see if indeed this wonderful photoshopped iPhone (I mean who would not want an iPhone running Windows XP) could be spotted anywhere else. And you guessed it: this little image is famous!
The iPhone running Windows XP is even on eBay for sale!
- Big Index! Getting Bigger…
Indices…That’s what dreams are made of! Our TinEye search index is approaching 500,000,000 images; that’s half a billion folks!
Killer Kitten on TinEye. View full size.
Half a billion may sound like a lot, doesn’t it? Well it is really just the beginning and that’s a tiny portion of all the images online today.
We are officially launching TinEye, our image identification based search engine, in a closed beta. What does this mean? You can request an invitation and servers permitting we will provide one! And you will get to play and experiment with the world’s first image identification search engine.
TinEye has been garnering great reviews already, from “the next frontier for web search” from the National Post to “indistiguishable from magic” from Jeff Barr, Amazon Evangelist to “image-recognition company that is second to none” Mathew Ingram.
You are all familiar with the Google or text based approach to searching: insert keywords in the search field and retrieve search results. TinEye uses an image as input instead. Rather than entering text, you upload an image to TinEye or you give it an image URL. It’s that simple! TinEye then uses our search index to retrieve where your search image has appeared in all the websites that we have indexed. No small feat.
And for your viewing eyes the TinEye video:
- Project Codename TinEye Launched in Private Beta
We not-so-quietly launched our internet-wide image search engine codenamed TinEye to our private beta testers today.
TinEye does for images what Google does for text.
Just as you are familiar with entering text in Google to find web pages that contain that text, using TinEye, you enter an image to find pages where that particular image (and modified versions of it) appears.
It’s a big step for us because our algorithms are now thousands of times more efficient than they were just a few years back. Uploading an image, and looking for matches in an index of over 487,000,000 images in real time is now a possibility. It’s something we’ve dreamed of doing for a long time, and now our beta testers are all over it.
Here are some of our favourite search results. The top image is the query image, and beneath it is the results.
- Everywhere Girl
I just stumbled upon Joey Coleman’s write up on MacLeans.ca about the Everywhere Girl and his reference to my blogpost about her. I wish I had seen his University of Manitoba blog post from 2005! Nice addition to our Everywhere Girl tracking. I haven’t seen her lately in our image identification reports since our book cover findings. I wonder if she has ever been used on a cover of a CD? We will soon find out!
- Uncle Same Wants You
We have been playing with our new image search service and doing all kinds of interesting searches. The questions in my mind is: when you do a Google image search, you typically find images that are tagged with your subject matter tag: say I am looking for “Uncle Sam” images, I find images that are tagged “Uncle Sam” or in a page with “Uncle Sam” text in close proximity to the image. Now how different would my image search results be if I used an image as the input. No tags, just an image of “Uncle Sam” and ask an image search engine to retrieve all of “Uncle Sam” images. How different would the results be?
I am betting they will be quite different. So we are continuing to play with our image search engine!
- Where can I find this lamp?
During my travels on the intertubes, I have a had a bad habit of collecting images I think are interesting. I drag images off of web pages and into my “unsorted” pictures folder which I check out every year or so for a laugh. The only thing that’s bad about this habit is that unlike a bookmark, there’s rarely context associated with an image file. There is rarely metadata about the images and filenames often provides no clues as to the source of the image. During my annual pictures folder perusal, I rediscovered this really inventive lamp design. I must have saved it from a web page but I don’t remember which one.
I don’t have any information about the image but would certainly like to find out more. Is this a real lamp I can buy? Who designed it? If it’s for sale, where can I buy it? There’s no metadata in the file itself and the filename wasn’t any help either.
How can I find what I want using only this image?
- Everywhere Girl, The Book
If you’ve ever worried about photos from your past coming back to haunt you, get to know the story of the Everywhere Girl. Over a decade ago she was a young actress posing for a series of stock photos. While she’s no Mona Lisa, in recent years her photos have made their way into royalty-free collections and crept into print and web designs the world over. First chronicled in Paul Hales’ technology blog The Inquirier and later by Idée’s own CEO Leila, the Everywhere Girl now even has her own blog. While fans have been compiling her images with the human eye for years, no method is better suited to this kind of task than image-recognition technology.
We indexed a series of Everywhere Girl photos using PixID our image recognition technology and have been monitoring the appearance of the Everywhere Girl. Here are the interesting results from our book cover monitoring project:
The Let’s Study Series of Christian books
- Peter (Let’s Study) by William W. Harrell
- Corinthians (Let’s Study) by David Jackman
- Galatians (Let’s Study) by Derek Thomas
- Top Mugs of 2007
2007 was a banner year for celebrity crime with eight mugshots breaking into the top 500 most used entertainment images of the year. Using the image recognition powers of PixID, we analyzed thousands of print magazines and newspapers to compile a list of the most used mugshots of 2007. Here they are in order of ascending popularity (least to most).
8. Actor Jason Wahler (The Hills) gets 30 days for alcohol related charges.
- Mona Lisa Gets Around
According to our PixID Image Monitoring Service, the Mona Lisa is one of the most used and abused images in print advertising today. Just using our image recognition technology, we compared millions of images against our growing collection of scanned print magazines and newspapers. The results were astounding – Mona Lisa really gets around.
Here are the most interesting sightings of Mona Lisa in advertising world:
PixID spots a tiny Mona Lisa in this Absolut advertisement