Recent results in 3D object recognition and 3D photography

Jean Ponce


I will first present some recent experimental results in 3D object modeling and recognition from photographs and video clips. This will include a comparison of our approach to 3D object recognition from photographs approach with several competing methods on a relatively small dataset (52 heavily cluttered images containing one to five instances of eight 3D objects taken from arbitrary viewpoint), as well as experiments in motion segmentation, scene modeling, and shot matching using shots from the "Run Lola, Run" and "Groundhog Day" films.


Maintained by Philippos Mordohai