VentureBeat

Google and Yahoo have image and video search, but they’ve kept things to themselves, oddly not syndicating these for other sites to use like they have with their text search engines.

No doubt, they will do so soon. Meantime, a host of specialized image search and recognition companies has sprouted up. Here are the latest ones:

Pixsy of San Francisco, may be the company to watch. They’re the only ones offering to license image and video search to any Web site, from large shopping sites to news sites. The logic for these sites is as follows: Why not have image search on your own Web site, instead of sending people on to Google? That way, the sites can enjoy advertising revenue (sharing some with Pixsy, which powers it), instead of give it all to Google. If a topic-based image search is wanted, Pixsy can do that too.

It has signed 50 different deals since launching its product in June, chief executive Chase Norlin tells us. For one, Pixsy powers Mamma, the Mark Cuban-backed search site that launched two weeks ago. Pixsy crawls the Web, and also crawling blog RSS feeds to get fresh user-generated images and video (Google has started doing this recently). Pixsy is generating revenue, and could continue to grow organically, but Norlin says he’s considering raising venture capital to make sure he can grab as much land as possible before Google hits. Today, Pixsy launched something called Pixsy Power, which lets any site use Pixsy’s search. It has 12 employees. Blinkx is Pixsy’s main competitor, but offers only video search.

Polar Rose, which we’ve mentioned before, has provided more details on its facial recognition technology.

The Malmo, Sweden-based company gave us an overview yesterday, but didn’t let us actually try it out — so proof will be in the pudding.

It appears to provide cutting-edge technology (based on creating 3D images from 2D images of photos), based on research at the University of Malmo. It mixes this with social networking — it looks for other photos around the Web with the same visual characteristics and that are tagged with similar names, thereby helping you detect who someone is. Polar Image has developed a pop up window that lets you tell it who it is, and to share and borrow from others using the technology (see image below.) The challenge will be for Polar Rose to find a real use beyond being a cool feature built into, say Apple or Adobe’s photo software.

It works by via Firefox and Internet Explorer plug-ins to be released early next year. It will also open its APIs for developers to work with its technology.

Chief executive Nikolaj Nyholm said he wants to develop similar technology for video. The company has raised $5.1 million from Nordic Venture Partners. Few other companies focus on facial recognition. Riya appears to have moved away from this area.

Ookles, the SF company that got funded in Om Malik’s loo, is apparently doing image recognition, but there are fewer details available on that (Techcrunch has some). Its founder described it eight months ago as a Flickr+Riya+YouTube, which sounded fun at the time, but eight months later, the idea is losing its freshness.

Finally, Snap just released its image search. And we mentioned Cognisign a few days ago.

Declaring it has “perfected the user experience” for audio and visual search, Seattle start-up Pluggd has raised $1.65 million from Intel and angel investors to help it start distributing its technology.

If you haven’t played with Pluggd, you should. It provides that “wow” experience, giving you what you intuitively want when searching video: a way to skip forward to the exact part of the audio or video file you are looking for. We’ll be hearing more about Pluggd next year, as it begins to cut partnership deals with major publishers, and comes out of the testing phase it launched two months ago.

Let’s take an example.

Take this ESPN radio recording from yesterday. Select the “find” tab, and type in the word “Marlins.” Pluggd will show you in the heat map the places most likely to be interesting to you. Orange shows a very high match. If you move the cursor there, you’ll hear the part about the Marlins. (You can do this by clicking on this image at left. You may be prompted to update your Flash player; go ahead and do so.)

But it gets even better.

Pluggd finds related words. Let’s say you’re looking for anything to do with injury, because you’d heard that Kobe Bryant might be injured. You type in “injury,” and Pluggd locates the part where the radio mentions his sprained ankle, even though the word “injury” is never mentioned in the audio. (Again, you can try this by clicking on image at left.)

This is impressive. Pluggd can do this by analyzing pages and pages of sports articles, and finding the statistical relationships between words. Its crawler finds that sprained ankle is very clearly correlated with the word injury over time. It does this without any sort of human domain experts. No one is doggedly typing in these associations behind the scenes. It is all automated, relying on the great database called the Web. “The Web itself represents mankind’s knowledge,” says Alexander Castro.

Right now, this cool search is only available at Pluggd’s demo site. And in case we’ve lost you, here’s a screencast tour.

Meanwhile, Pluggd has also building an inventory of ESPN and other files — now numbering more than a million — and it is busy indexing them all, so that it can make them available for crawling with its technology. Like Google, it wants to become a destination site. Also like Google, it wants to offer its technology to publishers, too, and Pluggd says it will be announcing various deals next year.

The company has boot-strapped itself until now, and the $1.65 million can be considered a seed round, to be converted into a first VC round sometime next year.

Intel made up a good portion of the investment, but more than half was contributed by a group of angels, including Scott Oki, former senior vice president for sales, marketing and service at Microsoft and Paul Maritz, former Microsoft group vice president of systems and applications. Other angels include:

–Brian Magierski, CEO of Kalivo, former co-founder/CEO of iMark (acquired by Ariba);
–Fraser Black, technology investor
–Bill Bryant, founder and investor in numerous search-focused startups including Netbot, Medio and Singingfish;
–Alex Alben, former executive at Starwave and RealNetworks;
–Barry Newman, venture partner at NeoCarta, former vice chairman of the technology group at Bear Stearns;
–Mark Klebanoff, former chief financial officer at RealNetworks.

There are a multitude of other companies focused on audio and video search (Pixsy, Podzinger and CastTV, for example), but none that are using Pluggd’s heat map approach that takes you directly to where you want to go.