With its modular infrastructure, it's easy to add new features to the system.


The system supports distributed computing for offline tasks using Apache Hadoop and index structuring optimizations for fast retrieval times.


The system is built using standard protocols and data exchange formats, making it easy to combine with other retrieval systems.

Core Components

The core of ParaDISE is made up of four basic components


Extracts local visual features from an image


Creates a vector that serves as a global representation of an image


Accesses visual indices (read/write operations)


Combines the results of multiple image queries or retrieval systems

Composite Components

These four basic components are used to create two composite components


Performs visual similarity search. Can use various distance and similarity metrics for the comparison of the vectors


Creates and accesses the visual indices. Supports building Approximate Nearest Neighbor search indices and weighting the indices


ParaDISE is currently used in the context of KHRESMOI, a European project a multilingual multimodal search and access system for biomedical information and documents


ParaDISE is also the backbone of Shambala, a modern and interactive image retrieval front-end which uses HTML5 and supports the Leap Motion™ motion sensor technology.

Google Glass™

ParaDISE serves as the backend of a search prototype developed for Google Glass™, which allows searching for images by using the device's integrated camera


ParaDISE is available in two editions, outlined below


This version of ParaDISE uses the Apache Software License 2.0. It provides alternatives to GPL-licensed libraries when possible, as well as MATLAB-based features. This version is suited for use in commercial applications (see licensing terms for more details).

ParaDISE-GPL (Coming Soon)

This version of ParaDISE uses the GNU GPLv3 License and includes various libraries which are licensed under the same terms, such as LIRE (Lucene Image Retrieval Engine), which provides implementations for several powerful visual features (CEDD, FCTH, ...). This version is destined for non-commercial applications.
In theory, there is no difference between
theory and practice. But, in practice, there is.
Jan L. A. van de Snepscheut


The recommended way of using ParaDISE is through the Maven build system. If you wish to start using the system without modifying the source code, download the pre-configured sample project (currently for the ASL edition only) and follow the instructions in the getting started guide.

Alternatively, you can add the following items to an existing Maven POM configuration file to get access to all the features of ParaDISE :

<!-- Access to the MedGIFT Maven repository -->
        <name>MedGIFT Maven Repository</name>
<!-- ParaDISE dependencies -->

Simply select the flavor of ParaDISE you want to use in the artifactId : paradise-{component}-asl OR paradise-{component}-gpl (coming soon)

NOTE :It is sufficient to add the Indexer and Seeker components to the POM file, as these will pull in the rest of the ParaDISE building blocks. You can also specify individual components as dependencies : paradise-extractor-{license}, paradise-descriptor-{license}, paradise-storer-{license}, paradise-fusor-{license}

If you wish to work with the source code directly, download the package containing all the Maven projects : ParaDISE-ASL OR ParaDISE-GPL (coming soon).

Once downloaded, unzip the archive and import the projects into your preferred IDE. Finally, install the Maven artifacts of ParaDISE into your local repository using the IDE or by executing the following Maven command in the root project's directory :

mvn install

Be sure to have a look at the guide as well, as it explains some key notions of the system

Download a sample set of images here. The link leads to a collection of >2000 images that can be used for an initial test of the system (indexing & retrieval). The collection is made available by infochimps, a company active in the fields of Big Data and Cloud Computing.

Getting Started Guide

This guide explains the few steps to set up the system and perform basic indexing and retrieval tasks.


You can access the Javadoc of ParaDISE-ASL here

Our Team

Here is a small overview of the ParaDISE development team.

Dimitrios Markonis

Lead Architect

Developed and structured the ParaDISE backend

Roger Schaer


Developed the text search component and the Web Service layer of ParaDISE

Alba García Seco de Herrera

Developer / System tester

Worked on the integration of visual features and real-world testing of the system

Henning Müller

Concept / Supervisor

Supervised and guided the development process of ParaDISE

Contact us

Got any questions, suggestions or other feedback? We'd be glad to hear your opinion and help you with any issues!

Simply contact us via e-mail: