Case Studies

Thanks to the extensibility principle, Ximilar is able to support a large range of applications, including:

Multimedia similarity search Ximilar technology can manage very large collections of multimedia data (images, video, audio) and search the collections according to various aspects of similarity. Scalability of this solution was proved by an on-line demonstration on 100 million images from Flickr.

  • finding similar e-shop commodities - let users in yor e-shop look for visually similar products
  • specialized image collections - visual similarity can be a useful function in many specialized image collections like trademarks, logotypes, or technical drawings
  • searching video databases - find video sequences with similar content that can be specified as any combination of visual, acoustic (sounds or speech) and textual (subtitles or annotations) information

Combination of several similarity aspects Various similarity aspects can be combined within a single search - for instance, visual similarity of images can be combined with annotation-based search to get result similar both visually and semantically.

  • similar photos in stock photography site - enrich the site with a capability to retrieve images with similar theme (see Pixmac)

Biometric data Majority of the biometric data types need to be searched according to similarity and Ximilar can provide high-performance database techniques to efficiently manage large collections of diverse biometric types.

  • person identification - online or offline identification of a person based on one or several biometric data types

Time series & other technical data The search engine can be used to efficiently manage a wide variety of scientific and other technical data.

  • currency rates or stock prices - search a database of stock price history

Copy detection, plagiarism, near-duplicates detection Since the engine can find similar objects very fast, it can be easily adopted to find copies or near-duplicates. For instance, applications that would like to fight plagiarism can employ the technology to check for the copied multimedia.

Image categorization The search technology of Ximilar can be used for categorization of images according to visual similarity.

  • disk scanning - scan a disk or other storage for images belonging to a category specified by a set of examples.

Local descriptors The local descriptors, available as one of the image descriptors in the engine, allows to search for similarities in subimages or help in object recognition.