People

Shenlong Wang, Sanja Fidler, Raquel Urtasun

Documents

Paper, Extended Abstract, Slides

Code

Coming soon.

Contact

For questions regarding the data please contact Shenlong Wang.

Relevant Publications

If you use the data and the code please cite the following publication:

• 

  Holistic 3D Scene Understanding from a Single Geo-tagged Image  (oral presentation)

  Shenlong Wang, Sanja Filder, Raquel Urtasun

  In Computer Vision and Pattern Recognition (CVPR), Boston, 2015

  @inproceedings{kitti3d,
  title = {Holistic 3D Scene Understanding from a Single Monocular Image},
  author = {Shenlong Wang, Sanja Filder, Raquel Urtasun},
  booktitle = {CVPR},
  year = {2015}}

  In this paper we are interested in exploiting geographic priors to help outdoor scene understanding. Towards this goal we propose a holistic approach that reasons jointly about 3D object detection, pose estimation, semantic segmentation as well as depth reconstruction from a single image. Our approach takes advantage of large-scale crowd-sourced maps to generate dense geographic, geometric and semantic priors by rendering the 3D world. We demonstrate the effectiveness of our holistic model on the challenging KITTI dataset, and show significant improvements over the baselines in all metrics and tasks.

  Abstract
  

  Bibtex
  

back to top