Volumetric dSfM Free Dataset
Introduction
The ground truth datasets consist of 20-frame mesh sequences of two articulated volumetric objects over time, with corresponding monocular image sequence. The monocular image sequence was generated by rendering the synthetically generated volumetric meshes from one camera perspective. The real-world dataset consists of a monocular image sequence depicting an articulated jointed doll together with a 3D template geometry of the initial state.
Download
Synthetic data
|
Pixarlamp sequence
pixarlamp.zip |
|
|
Sackboy sequence
sackboy.zip |
|
Real data
|
Puppet sequence
puppet.zip |
|
Requesting the data
All submitted papers or any publicly available text using the volumetric dSfM Free Dataset must cite the following paper:
Related publication
Kausch L., Hilsmann A. and Eisert P., Template-based 3D Non-Rigid Shape Estimation from Monocular Image Sequences, Proceedings of the 22nd Workshop on Vision, Modelling, and Visualization (VMV), Bonn, Germany, September 2017