imprezzeo

Background information for media

Imprezzeo is an image-based search technology company. Our technology allows users to search for images using other images as examples rather than textual search terms.

 

The beta launch of Imprezzeo is Q4 2008.

 

 

What does Imprezzeo do?

Imprezzeo provides users with quicker, more relevant image search results by using images to search for images, rather than text. The Imprezzeo software is based on proprietary content-based image retrieval (CBIR) and facial recognition technologies and generates far better results because it identifies images that closely match a sample set of single or multiple reference images – unlike keyword-based searches that often return many unrelated images.

 

 

How can it be used?

Imprezzeo’s technology can be used in any number of different applications. The primary application is image search – searching for images held in an online digital photo library, an internally-used digital collection, an online photo-sharing site or an internet search engine.

 

Currently, when a user searches a collection of images for a particular image, they are often presented with an unmanageable number of pages to trawl through in order to find the right picture. This is because the collections they are searching are often very large, the keywording imprecise and user search terms ‘simple’. Using Imprezzeo, the user can refine an initial results set by selecting a sample set that resembles what they are looking for. Alternatively, the user can upload their own sample set using images from their own collection. Either way, Imprezzeo analyses the actual content of the sample set and retrieves images that look like those within it.

 

Beyond image search and beyond the enterprise

Imprezzeo's current focus is the enterprise, particularly media - stock photo libraries, news agencies, publishers. We do however see huge scope for our technology in other markets and in other applications; for example in web image search, web photo-hosting, sharing and finishing, retail and many more. We are currently formulating our plans for this next phase.

 

 

What’s wrong with the current search methods?

Text-based search systems have many disadvantages:

• Images are very subjective. Two people might interpret two very different things from an image and could subsequently label the image differently If an image is mislabelled or the keyword terms are ambiguous, the search engine will often return irrelevant results

• Tagging guidelines change regularly and are different for each company, which leads to inconsistency and potentially a negative experience for the end user

• Keyword-based approaches to search presume users have a clear idea of what they’re looking for and are able to articulate that in their search terms

• Keyword-based approaches to search are costly to maintain. So much so that for most image libraries it’s not economical to tag all material, which results in missed revenue opportunities

 

Who uses Imprezzeo?

Imprezzeo is the perfect complementary technology for search engines; and it is ideal for any organization with an extensive library of digital images, such as stock libraries, news agencies and publishers. It also enhances the appeal of any other companies where users regularly conduct image searches, such as retail outlets, dating and photo-sharing sites.

 

 

Benefits to customers

Imprezzeo delivers tangible benefits as a search application. For end users, it conducts a quicker and more accurate image search. From a commercial perspective, the vastly improved search facility afforded by Imprezzeo means that companies can increase sales from existing and future images because they can present the most relevant range of pictures to customers. And because all of the images returned are relevant to the search, the customer is far less likely to be dissatisfied and drop off the site, and instead will be more inclined to make return visits and purchases from the site.

 

There are also potential cost savings in reducing the reliance on keywords. Most large image collections rely on images to be richly tagged so that they might be returned when one or more of those keywords feature in the user’s search terms. This is a costly, labor-intensive and inexact process. By letting the user provide examples of what they are looking for, the extent and accuracy of keywording becomes much less of an issue, resulting in a significant reduction in the cost of image processing.

 

Using Imprezzeo as an image detection application can also help organizations generate additional revenue by ensuring that all copyrighted images used online are duly paid for.

 

 

Will existing image search software need to be upgraded or replaced?

No, Imprezzeo is a complementary solution that integrates seamlessly with existing search technologies, and even allows businesses to continue using their existing user interfaces.

 

Crucially, end users search in exactly the same way that they’re accustomed to – they simply select a sample set that looks like the images they require, rather than having to specify more and more text descriptions.

 

Imprezzeo easily integrates with most digital asset management (DAM) and content management systems, while maintaining existing search tools and user interfaces. Specifically, Imprezzeo supports SOAP (Simple Object Access Protocol) web interfaces, which allow Imprezzeo core processes to be exposed as web services, thereby reducing the cost and complexity of content or application sharing.

 

 

The Imprezzeo software

Imprezzeo uses two forms of technology: CBIR and facial recognition. These technologies have been developed by two Australian universities – University of Queensland and the University of Wollongong. The Imprezzeo development team have highly successful research and development backgrounds, some with more than 20 years’ experience, and are experts in CBIR and facial recognition technology.

 

CBIR works by analyzing the actual contents and attributes of the sample set, including the color definitions, spatial layout, shapes, texture and facial information. Once the image has been analyzed, the contents will be used to search for any matching images stored in a given database.

 

The robust Imprezzeo technology is built on Java and C++ and has been designed to integrate quickly and easily with all existing IT systems and image libraries.

 

 

How is Imprezzeo different from other image-based search companies?

Imprezzeo’s solution is unique and superior because:

• It looks not only at color information in any image but a much broader set of parameters

• It combines CBIR and facial recognition techniques. This innovative fusion of two different technologies means that Imprezzeo can provide additional face detection and face recognition features that are unique in image search, allowing users to conduct accurate searches for images with people in them, or images with specific persons

• It allows the user to select one or more pictures as sample images

• It can scale to work across image collections ranging from the thousands to the many millions – with no adverse impact on processing speeds

• The user can continually refine their result set by selecting further sample images

• It can be customized to perform optimally with a specific collection

 

The company

Imprezzeo is a dynamic, ambitious and well-backed early stage technology business. Seed funding and further development has been provided by Independent News & Media since 2007.

 

The company’s management team includes CEO, Dermot Corrigan, who has worked previously for Frost & Sullivan; LexisNexis; PR Newswire; and Wolters Kluwer. Imprezzeo’s CTO, Peter Chin, has worked for Kaz (part of Telstra); and Logica CMG. Kurt Dressel, Imprezzeo’s Vice President, Sales, has worked for Vio Worldwide; Pindar Systems; Quark, Inc; and Gannett Media Technologies.

 

Imprezzeo has offices in London, New York and Sydney.