ROBIN Competition

Datasets, Ground Truths and Metrics
For the evaluation of
object recognition and image categorisation


The ROBIN project aims to produce datasets, ground truths, competitions and metrics for the evaluation of object recognition algorithms. The scope of this project includes :

  • multi-class objects detection
  • generic objects detection
  • generic objects recognition
  • image categorisation

This project is funded by the french ministry of defense, the french ministry of research (Techno-Vision funds) as well as by several companies and research centers (Bertin technologies, CNES,Cybernetix,DGA,EADS,INRIA, ONERA, MBDA,SAGEM, THALES).

The production of the resources will be completed on february 2006. Competitions and benchmarking will follow. Robin is planned for two years.

The production of the ressources is almost finished. Discussions are concerning the metrics and competitions. Competitors are welcomed at any time.

Go to the Contact Informations Page and see how to be informed and how to contribute to this project.

The final proposal can be downloaded here (french document) : [[PDF].