3DVis dataset includes a set of 12 heterogeneous scenes for testing 3D scene registration and analysis methods. Models include homogeneous shapes, repetitive photommetric patterns and symmetries. Matlab © scripts for data handling are included. Scenes are stored in Wavefront .obj format as point clouds with vertex and normals information
3DVis is primarily MPL2 licensed (FAQ). Some files contain third-party code under other licenses.
If you use 3DVis dataset in your academic projects, please cite the papers we implement as appropriate:
@article{cirujeda20153d,
title={A 3d scene registration method via covariance descriptors and an evolutionary stable strategy game theory solver},
author={Cirujeda, Pol and Cid, Yashin Dicente and Mateo, Xavier and Binefa, Xavier},
journal={International Journal of Computer Vision},
volume={115},
number={3},
pages={306--329},
year={2015},
publisher={Springer}
}
@inproceedings{cirujeda2014mcov,
title={MCOV: a covariance descriptor for fusion of texture and shape features in 3D point clouds},
author={Cirujeda, Pol and Mateo, Xavier and Dicente, Yashin and Binefa, Xavier},
booktitle={2014 2nd International Conference on 3D Vision},
volume={1},
pages={551--558},
year={2014},
organization={IEEE}
}