Mondays on AltSearchEngines we examine a particular Vertical, and this week the hottest Vertical category is Image Search, and especially the sub-category of facial recognition search engines.

Today was the partial launch of EyeAlike, which matches up faces with similar faces.  It works is two directions.  On the top level, you supply the image of someone that you find attractive, and EyeAlike scans thousands of faces from Online Dating sites, and gives you a real person, someone seeking a date, that is very similar to the photo you uploaded.  This is meant to be a serious endeavour.

One the bottom row, it is just the opposite, you supply the face of a real person, perhaps your own, and EyeAlike scans thousand of celebrities and shows you which celebrity that person looks like!  This is really just meant to be a novelty.

To access the site, you will have to request a password as long as it’s in Private Beta.

 Mugr is another project getting a bit of attention lately. 

“Our face detection is surprisingly good, but recognition in the face of arbitrary background noise is a difficult challenge.”

“In case anyone’s reading and concerned about their privacy, let me be very clear about certain things: 
Mugr is an opt-in service.”

“If you don’t sign up, we can’t find you. Period.  Face recognition does not work without photos.  If you sign up and don’t upload a picture, we can’t find you in photos.  We find you using your name as well as through your tags and links, but not via your photo.”

“If you upload photos, add tags and links, and make connections with people these are the things that you cannot set to private. Everything else, you have complete control over. Of course, you’ve got control over tags, links, and photos as well— you can choose to add them or not.  We will not be spidering photos from the Internet to seed our searches.” 

As for other new Facial Search engines,  one would be Viewdle, a facial recognition search engine that we covered recently.  Viewdle claims to be able to search for faces within a video stream.

Viewdle is a facial-recognition digital media platform for indexing and searching videos. Viewdle automatically looks inside the video, frame-by-frame, to create a real-time index of true on-screen appearances.

If you want to experiment with Viewdle, you should go to the REUTERS Labs test site.

Since I can’t wait to see her movie, The Brave One, I searched for actress Jodie Foster.

And here we are, Viewdle’s Facial Recognition Search Engine:

PolarRose looks for still images that have been tagged by other PolarRose users. The Polar Rose browser plugin for FireFox lets you discover who’s in any public photo. The browser plugin is currently in private beta so you need to sign upto be notified. The plugin adds their signature rose on all public photos where it detects a face.

Pluggd looks for words and images in audio and video files by their heat map signatures. SeeHere uses an intuitive heat map interface to let you search inside a video and jump right to the parts that interest you.




“Imagery has only appeared so because it’s new reincarnation, Domburi, is undergoing labor pains. I’m starting afresh, new name, new domain, new engine. After all, Imagery was always meant as a mere proof-of-concept and it was only with its onslaught reception that I realized I was on to something.



Finally, there’s Facesearch, one of the first facial searches (it crops out everything but the face).

This entry was posted on Monday, November 12th, 2007 at 8:55 pm and is filed under In Beta, Innovations, Verticals. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.