pyCovariance is a python package for statistical estimation and clustering/classification on Riemannian manifolds.
It implements the following pipeline (where M is a Riemannian manifold):
This pipeline can be applied to many other types of data than images such as time-series.
The script install.sh creates a conda environment with everything needed to run the examples of this repo and installs the package:
./install.sh
To check the installation, activate the created conda environment pyCovariance and run the unit tests:
conda activate pyCovariance
nose2 -v --with-coverage && coverage combine
To run examples, run the scripts from the folder examples/ e.g.
python examples/hyperspectral/demo.py
If you use this code please cite one of the following papers:
@misc{collas22MSG,
author = {Collas, Antoine and Breloy, Arnaud and Ren, Chengfang and Ginolhac, Guillaume and Ovarlez, Jean-Philippe},
title = {Riemannian optimization for non-centered mixture of scaled Gaussian distributions},
year = {2022},
url = {https://arxiv.org/abs/2209.03315}
}
@ARTICLE{collas2021ppca,
author = {Collas, Antoine and Bouchard, Florent and Breloy, Arnaud and Ginolhac, Guillaume and Ren, Chengfang and Ovarlez, Jean-Philippe},
journal = {IEEE Transactions on Signal Processing},
title = {Probabilistic PCA From Heteroscedastic Signals: Geometric Framework and Application to Clustering},
year = {2021},
volume = {69},
number = {},
pages = {6546-6560},
doi = {10.1109/TSP.2021.3130997}
}
@INPROCEEDINGS{collas21tylertype,
author={Collas, A. and Bouchard, F. and Breloy, A. and Ren, C. and Ginolhac, G. and Ovarlez, J.-P.},
booktitle={ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
title={A Tyler-Type Estimator of Location and Scatter Leveraging Riemannian Optimization},
year={2021},
volume={},
number={},
pages={5160-5164},
doi={10.1109/ICASSP39728.2021.9414974}
}