Name (required):

Email Address (required):

Please state your business or university affiliation (optional):

Please describe your research purpose (optional):

2ppl sculpture dataset (UNC Chapel Hill)

Due to the huge size of the files (appx. 1.8GB in total), I have zipped it into 10 files as follows. Sorry for the inconvenience.
(1) projection matrices of the 9 camera views
(2-10) un-conpressed avi videos for background training and the actual sequence, see detailed description below

In the dataset there are 9 avi videos of camera views of 720*480 frames. The video is resampled from the original 29.99 Hz video with the sampling framerate at 3Hz. And then the frames are re-compile into uncompressed avi at 29.99Hz. The first 30 frames are background images without dynamic objects (people) in the scene. The rest 215 frames are synchronized over cameras for papers

• Li Guan, Jean-Sebastien Franco and Marc Pollefeys, "3D occlussion inference from silhouette cues", CVPR 2007.

• Li Guan, Jean-Sebastien Franco and Marc Pollefeys, "Multi-Object Shape Estimation and Tracking from Silhouette Cues", CVPR 2008.
• Li Guan, Jean-Sebastien Franco and Marc Pollefeys, "Probabilistic Multi-view Dynamic Scene Reconstruction and Occlusion Reasoning from Silhouette Cues", IJCV 2010.

Please contact Li Guan at lguan@cs.unc.edu for questions regarding this dataset. Although I have left UNC, I can still be reached from this email address.
Version 0.0 May 2007