Information Today, Inc. Corporate Site KMWorld CRM Media Streaming Media Faulkner Speech Technology Unisphere/DBTA
Other ITI Websites
American Library Directory Boardwalk Empire Database Trends and Applications DestinationCRM Faulkner Information Services Fulltext Sources Online InfoToday Europe KMWorld Literary Market Place Plexus Publishing Smart Customer Service Speech Technology Streaming Media Streaming Media Europe Streaming Media Producer Unisphere Research

News & Events > NewsBreaks
Back Index Forward
Threads bluesky LinkedIn FaceBook RSS Feed

Imprezzeo Offers a New Way to Search for Images
Posted On December 15, 2008
While search has become a major industry over the past several years, one corner of the market that has largely been relegated to the sidelines is image search. Imprezzeo ( is an upstart company whose focus is filling in the gaps in image search where other search technology companies have lagged.

"Over the past several years, we’ve witnessed a massive proliferation of images both online and in the enterprise," says Dermot Corrigan, CEO of Imprezzeo. "People struggle every day to sift through all those images to find the results that they want. But we’re still applying Web 1.0 solutions to a Web 2.0 problem."

Imprezzeo’s solution is a software tool that uses proprietary content-based image retrieval (CBIR) and facial-recognition technologies to generate results sets for images. As opposed to standard search engines such as Google, which utilize text-based searching for images, Imprezzeo’s technologies employ analyses of the images themselves when returning results.

Imprezzeo launched its solution in beta earlier this month. Right off the bat, the company believes its product will be most valuable to organizations with extensive libraries of digital images, such as stock libraries, news agencies, and publishers.

In the course of doing business, a newspaper, for instance, may need to locate a photo from its archive using information that is not easily translated into metadata or a text-search query. Imprezzeo’s technology allows a user at the organization to use a sample image set or to upload his or her own image to find archived images that match characteristics such as color, definition, spatial layout, shapes, texture, and facial information.

"We’ve analyzed the market quite clearly and tried to understand where most ROI is achieved," Corrigan says. "Stock photo libraries may get millions of hits every year. Their worry is that their users are struggling to find the content that they need, even though there’s a high likelihood that they have the image that they want. Their options are to employ a keyword search and hope that every image has been keyworded correctly—or, far better than that, they can let their user provide an example and let Imprezzeo take over."

Imprezzeo has identified a number of weaknesses in keyword search that illustrate the superiority of image-based search. Any human analysis of an image is bound to be subjective; images may be mislabeled, labeled ambiguously, or suffer from any number of other labeling errors. Imprezzeo’s technologies apply one set of analytical standards, allowing users to obtain more focused search results.

Other problems of text-based search for images that Imprezzeo points to include rapidly changing tagging guidelines; the presumption that users are able to clearly articulate what they’re looking for when searching for an image; and the cost of maintenance of tagging huge image archives.

"What you end up with is, instead of a user sitting there for hours on end, scrolling through screen after screen, they can provide an example of the image they’re looking to match and get relevant results and after one or two clicks," Corrigan said.

Corrigan says that what separates Imprezzeo’s image-search software from the software of competitors in the field is its use of content-based image retrieval as opposed to strictly facial-recognition software. A recent article in The New York Times ("Zeroing In on Your Favorite Video Clips," Dec. 5, 2008) described how companies such as VideoSurf and Digitalsmiths search online videos using facial-recognition software.

Facial-recognition technology is good, Corrigan says, for searches aimed at making exact matches of images for purposes such as copyright protection. "That’s not our thrust," he says. "Similarity is where we come at it. Our solution allows users to find images that are similar to one another, not necessarily identical."

As Imprezzeo begins to branch out from the enterprise to the mainstream, Corrigan envisions numerous commercial applications for the technology. Photo sites, such as Flickr, would be able to offer search-by-image rather than relying on users to tag their own photos; consumers would be able to automatically organize their photo collections by subject rather than by date; and retailers would be able to use the technology to offer similar products to online shoppers.

Imprezzeo maintains offices in London, New York, and Sydney, Australia. The company’s technologies were developed by two Australian universities: the University of Queensland and the University of Wollongong. Corrigan was appointed CEO of Imprezzeo in 2007 after having worked at Frost & Sullivan, PR Newswire, and LexisNexis.

The startup is backed by Independent News & Media PLC (INM; Imprezzeo got its initial seed and further development capital investment from INM in June 2007 and has spent the past year refining its software applications and patents.

Gavin O’Reilly, group COO, INM, comments: "Imprezzeo is simply unlike any other image-based search proposition in the market today, because it is based on two cutting-edge technologies, guaranteeing relevant image search results each and every time. Using images to search for images is what the market is looking for, and we see Imprezzeo as a core complementary technology for search engines that will transform the way users search for images."

Michael LoPresti is the former assistant editor of EContent magazine. He is currently a graduate student and freelance writer living in Syracuse, N.Y.

Email Michael LoPresti

Related Articles

2/2/2021Exploring the Data Around Facial Recognition

Comments Add A Comment

              Back to top