 | LifeCLEF: The new lab of CLEF dedicated to multimedia life species identification. In the continuity of the image-based plant identification task of Imageclef, this new system-oriented benchmark considers birds and fish in addition to plants, audio and video contents in addition to images, scales-up the to hundreds of thousands of life media records and thousands of living species. |
 | ImageCLEF Plant Id task: A benchmarking task and dataset aimed at evaluating leaf-based plant identification systems in collaborative image collections. It was set up within ImageCLEF 2011 2011 and extended during the 2012 and 2013 campaigns. |
 | FlickrBelgaLogos: A new challenging dataset set up in 2012 and aimed at evaluating small objects discovery and mining methods in large natural image collections. It was created in the scope of the last publication of my phD student Pierre Letessier LetessierACM-MM2012.pdf |
 | BelgaLogos: A 10K images dataset set up in 2009 and aimed at evaluating logos retrieval methods in large natural image collections. It has been enriched in 2012 with a larger groundtruth and precise localization information (more information on the page). |
 | TRECVID 2008 Video Copy Detection. I was co-organizing this first TRECVID task dedicated to the evaluation of Content-Based Video Copy Detection methods. The data are not freely available but you can download the tool we developed to create attacked videos. It was still used by NIST for 2009 and 2010 campaigns. |
 | MUSCLE/CIVR Video Copy Detection was the first benchmark dedicated to the evaluation of Content-Based Video Copy Detection methods. The MUSCLE-VCD-2007 corpus was created for this event is is freely available. |