The tail end of storm clouds over Santa Barbara, at the end of a rare storm in  paradise - David Boardman 2003
 
TECHNOLOGIES
. . . Image search and categorization based on visual clues
 
       

VIMA Search

 

``The perception-based technologies recently developed by VIMA have made marked strides in enhancing the accuracy and speed on multimedia information retrieval. In particular, VIMA's inventions in learning algorithms, perceptual distance functions, and indexing methods have given the company significant lead over the other players in the field.´´

Dr. Yong Rui
Scientist
Microsoft Research

VIMA Search

VIMA Search is software that opens up a new way for people to search for and manage multimedia content visually.  Based on our patented perception-based image retrieval (PBIR) core technology, VIMA Search allows users to find images from large image collections in just a few seconds by clicking on sample images dynamically suggested by the system.  Using a multi-modality approach that incorporates complex image features, textual hints, and 'learning' algorithms, VIMA Search offers the most accurate and efficient image search engine on the market.  Its innovative approach to image matching can be leveraged across several market categories, including image search, classification, and recognition.

Increase Revenues
Industry research reveals that approximately $13 billion was lost by online businesses last year, due to frustrated customers not being able to find what they were looking for. VIMA Search can help to eliminate this problem for your company by giving your customers a way to search for your products quickly, easily and accurately. A better, faster search results in more sales and higher revenues.

Boost Customer Satisfaction
Aside from increasing your sales-per-search ratio, installing VIMA Search will help keep your users satisfied and returning to your site. By allowing a new, effective way for people to find what they're looking for, VIMA eliminates the frustration created by the limitations of traditional search capabilities. And, if your customers are pleased, they'll provide you with good references and referrals to help your online business grow..

Provide a Personalized User Experience
VIMA Search uses both positive and negative feedback from a user to learn his/her preferences. Because each person is looking for something different, the search engine will provide each with a different set of results that best match his/her criteria - something other companies can't do.


VIMA Search Features

Easy Installation and Maintenance
VIMA Search's high-dimensional indexer allows for quick, simple system integration and expansion.

  • Is easily applied to existing image databases; indexes pictures automatically
  • Multimodal capability allows seamless incorporation with current text-search functionality
  • New images can be added to the system within seconds
  • Easily scalable for increases in image database size and complexity

Highly Accurate
Use of Perception-Based Image Characterization makes VIMA Search the highest-quality image search engine available.

  • Patented technologies use artificial intelligence to analyze 150+ parameters for faster and more precise results
  • Extensive feature database makes for more accurate matching of images to preferences
  • Uses cognitive psychology theories to make pairing more "natural and intuitive," closer to what users want

Adaptable and Efficient
Revolutionary intelligent learning algorithms and sampling process enable VIMA Search to understand users' target image characteristics quickly and easily.

  • Requires minimum number of iterations (user feedback) to return accurate results
  • Can pinpoint user preferences quickly - no need to specify text or select particular features (color, texture, etc) - just point and click
  • Easily handles changes in concept preferences and/or complexity - even in mid-search

Customizable
Multimodal query support means VIMA Search is appropriate for all image databases, regardless of specifics.

  • Handles three different seeding modes for search initiation - seed-by-keyword, seed-by-content, seed-by-nothing
  • Operates on several levels, including text-based, content-based, or a combination of both
  • Can be adapted to handle different languages for easy international application