These datasets are made available for research purposes. In some cases, to obtain a representative dataset, we have obtained images using a Google image search without permission from the original copyright holders. By downloading these files, you agree not to hold the authors or Carnegie Mellon University liable for any damage, lawsuits, or other loss resulting from the possession or use of these images and/or ground truth files; you also acknowledge that this dataset may change at any time without notice. If you are the copyright owner of one of these images and would like it removed from the dataset, please contact Derek Hoiem.

Automatic Photo Pop-up

Included are the images used for training and testing our Automatic Photo Pop-up (SIGGRAPH 2005) system and their ground-truth labels (in .mat files). See the included README for more information.
Ground Truth

Surface Context

Included are the images used for training and testing our geometric context(ICCV 2005, IJCV 2007) systems and their ground-truth labels (in .mat files). See included README for contents and copyright information.
Geometric Context Dataset

Putting Objects in Perspective

This dataset is a subset of the LabelMe database. See README for description.
Test Images and Ground Truth
Validation Images and Ground Truth


Some of this material is based upon work supported by the National Science Foundation under CAREER Grant No. ISS-0546547. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. This work is also partially supported by a Microsoft Research Fellowship awarded in 2006.