Google is always experimenting. That’s a fact. Since purchasing Riya back in November of 2005, everyone’s been expecting to see a facial recognition refinement offered on the advanced image search page. While it hasn’t been officially added, word has been passed around from bloggers and news sites about Google’s new image results trick.
Google’s Latest Hidden Image Search Refinements
Turns out that Google does have the ability to refine image searches to include facial recognition factors, but you have to know how to do it at the base URL since there’s no live button for it on the site. You can limit the search results to include human faces by typing &imgtype=face at the end of the URL string. For example, an image search for the word gates produces a page of images including jubilee gates, palace gates, Bill Gates, castle gates and even a photo of one of Bill Gates’ cars:

But add &imgtype=face to the end of the URL string and it restricts the image results to include human faces, showing 12 pictures of Bill Gates along with a few photos of other people including Skip Gates, Kent Gates and Robert Gates.

It’s also possible to restrict your image searches to news items, in the same URL-manipulating way. At the end of the URL string on an image search place the string &imgtype=news to see images from sites that Google considers news sources:

Pretty neat stuff, huh? Google is often credited with cool new tricks and innovative offerings. The buzz around their new image search refinements prompted me to take another look at what’s happening on the rest of the search engine landscape at large. Here’s what I learned:
Yahoo Image Search
If you search for gates on Yahoo’s image search, the results page returns images of gates, the wrought iron kind, with Bill Gates appearing once at the bottom of the page in the next to last image (a flickr.com photo of course). However, Yahoo offers the chance to refine your search right at the top of the page with the Also try: links:

Yahoo’s advanced image search offers nothing in the way of filtering for face recognition. Making additions to the URL strings to tweak what Yahoo adds for large image restrictions and color image restrictions finds no little ‘trick’ capable of returning other restrictions than what’s already offered. I guess if they had one it would have been passed around the blogosphere by now.
MSN Image Search
Amusingly enough, searching for gates using MSN’s image search will give you a results page chock full of photos of Bill Gates. Okay, there is one picture of a wrought iron gate and a few pictures of what are said to be Bill Gates’ houses:

The one thing that MSN offers that other image searches don’t is their scratchpad. Off to the right-hand side of the page you can open and close the scratchpad, dropping and dragging photos as you please.

It’s pretty handy if you want to compare images as you go along or build a collection to look at more closely later. Although I can’t imagine what else I might want to do with a selected collection, it is a neat feature that the other image search sites don’t offer.
ASK Image Search
Ask Images produces a fair sample of both people and fixtures for the gates search. What’s interesting to note here is that they already show the news images separated from the web images, without the need for typing extra strings to the URL. While they don’t offer a filter for face recognition - they do offer search refinement choices on the right that includes a group selection for related names, as well as the expand and narrow your search options:

Related Names will give you results that include people, with faces, but the images are still obviously refined through contextual clues.

Exalead Image Search
Exalead.com, the self-proclaimed other search engine is a lesser known but highly functional and user-friendly search engine that actually offers the best image search capabilities on the Internet today. Exalead was founded in 2000 as an enterprise search company, offering businesses a way to access information within organizations of all sizes. It’s touted as easy to deploy and administer, adapting to user habits for a unique search experience. When you search for gates on Exalead’s image search, you receive a mixture of gates, from expanding shutter gates used in retail stores to gates at Graceland and Bill Gates. But look, there’s already a way to narrow your search by image content and the single selection offered is face.

Clicking on face to narrow your search gives you photos of Bill Gates, Ronald Gates and a few portraits of other people:


The narrow your search panel on Exalead offers more function and user-friendly features than other search sites do. Click on more choices to see the entire list of options, including size, content, wallpapers, color, grayscale, black and white, layout and file types. What makes it even more user friendly is that as you sift through the options, you can go back and remove any filter you’ve turned on along the way. I managed to find a nice little portrait of Bill Gates using these search refinements. Had I changed my mind and chosen to go with a grayscale photo instead, I could simply remove the image color filter and change it to find the grayscale images:

As you drill down in the choices and options, Exalead tells you how many results are within each of them. For example, if you decide that you don’t want a color image in the above search, removing it will give you the option of going back to choose something else. The image color choice option now tells you that for an
image search on gates, with the content defined as face, the file type of Jpeg, in small size with the portrait layout and bill as the search within results refinement. There are 328 color images available and 30 grayscale images available. I honestly can’t think of anything easier than that. Select Grayscale instead and get the following results page:

Exalead’s advanced search offers a pop up on top of the existing page that allows you to set parameters quickly and easily with helpful examples provided. As you click on the options, the refinements are placed in the box above and you simply add in your parameters, then click on the button and find what you’re looking for.

In all fairness… I wouldn’t have known about Exalead’s facial recognition capabilities had I not read so much about Google’s hidden feature over the past week. I admit that I am as Google-centric as everyone else when it comes to finding things online. I check the other search engines occasionally but I start with Google and usually manage to find what I’m looking for there. So I took the time to explore Exalead.com again since I had been remiss in doing so over the past year or so. As it turns out they are really offering a search resource that outperforms many of the mainstream search engines in many ways. They are also sporting a new blog, since March, that should help keep me more informed about what’s going on over at the other search engine. In a posting on April, 23, 2007, they formally announced their facial recognition capabilities going live:
Our teams had the pleasure of collaborating with LTU Technologies, a company specializing in image recognition and searching, on one of the most exciting new features, search refinement by content type, commencing with a filter for faces. Exalead Becomes A Facial Recognition Expert; April 23, 2007
And just last week Nicholas, Web Project Manager, discussed Google’s Universal Search in comparison to their Search By Serendipity which, according to Nicholas…
has been at the very heart of Exalead’s technology since day one : between different vertical search engines through the little “Exa mascot” that suggests audio or video results in the Web search, and in the more unique way within each vertical Exalead search application using a tag cloud of concepts to navigate in the search results (of our Wikipedia search for instance).
At the end of the posting, Nicholas pushes the bar and questions the timing of Google’s image search refinement trick:
On a similar track, we are very pleased to see that Exalead’s latest innovations (image search with face recognition) has also been introduced (although still in a hidden form) a few days ago by Google. Maybe they had to release this ahead of schedule? Who knows… Imitation is the sincerest form of flattery; May 31, 2007
The Image Search of Choice After conducting this image search comparison project, I realize that it’s time for me to make a change; at least for image searching. Exalead’s Image Search, by far, offers the most advanced search capabilities than any of the others I compared. When I know exactly what kind of image I’m looking for, it will be the first place I go.
Web Ad.vantage is a full-service online marketing company with core competencies in search engine optimizatiom, PPC Campaign Management and online media buying. Visit our Internet Marketing Services section to learn more about our full range of services.
WebAdvantage.net encourages the reprinting of our marketing tips and articles. Before doing so, however, please contact us at for permission to do so. The company bio located above is required to accompany any reprint. Thank you in advance for your professional courtesy.
Pragmatic, professional advice with no hidden agenda.
![]()
Internet Business Forum
Find out more hereOlympics Ad Spend Not Quite Gold Medal Worthy
Online researchers eMarketer released data on August 22nd that estimated NBC’s Olympics video advertising spend at 5.75million. The Olympics has brought record numbers of site visitors to NBCOlympics.com as well as TV viewers to the network... read more
U.S. Women and the Internet, Part 1
This article by Hollis originally appeared in ClickZ on February 26, 2008. What do women want? Women’s use of the Internet and their online presence is huge, yet I feel the interactive advertising industry has treated online women as... read more
Cuil: Cool or Uncool?
It seems that everybody’s “Googling” these days—but are you “Cuiling” yet? Cuil (pronounced “cool”) is a new search engine developed by former Google engineer and search architect Anna Patterson and her husband Tom Costello (a former... read more















back to top
Subscribe to our blog RSS



