Accueil

Welcome to the beta version of our 3D research engine website: 3Dretrieval!

In this web site, you will be able to retrieve 3D models from a photo or by choosing a 3D model from the database. If you are not yet a member, you can create an account by clicking on "Create your account". This will allow you to have your own images historic.

You can use the search engine by loading a picture from your computer or from the collective historic by clicking on "Start 3D retrieval". Once the image selected, click on "Load the picture" then "Start the 2D search". For more accurate results, you can select one of the resulting 3D model and click "Restart the search on this 3D model" button.

The search engine has been created and developped by Tarik Filali Ansary during his Ph.D. thesis (2003-2006) under the supervision of Mohamed Daoudi and Dr. Jean-Philippe Vandeborre. This work has been supported by the RNRT SEMANTIC-3D project (scientific responsable: Mohamed Daoudi).

More details can be found in A Bayesian 3D Search Engine using Adaptive Views Clustering, published in IEEE Transactions on Multimedia, volume 9, number 1, pp. 78-88, January 2007 [PDF].

The 3D Model Database used in this search engine is the Princeton Shape Benchmark. This database contains 1814 3D VRML models.

For any information mail to: Mohamed Daoudi or Jean-Philippe Vandeborre.

Enjoy!