Overview

FIRE, the Flexible Image Retrieval Engine, is an image retrieval system designed for research in this area. It is focus on evaluating different image descriptors, not on indices for efficient search.

It was mainly developed by Thomas Deselaers during his diploma and Phd Theses at the Human Language Technology and Pattern Recognition Group at RWTH Aachen University in Aachen, Germany.

Contributors

• Thomas Deselaers
• Daniel Keysers (initial version/lots of help/...)
• Andre Hegerath (sparse histograms, ...)
• Tobias Weyand (combination with text retrieval, ...)
• Helga Velroyen (reimplementation and debugging of some feature extractors, ...)
• Jens Forster (on demand feature loading)
• Christian Terboven (openMP parallelization)
• Fabian Schwahn (upload of query images and querying with subimages)
• Christian Kofler (FireWatch)
• Piotr Boninski (C# library to access a FIRE server)
• Tobias Gass (diversity reranking, some smaller bugfixes, svm scoring)