SPDNet-master

Zhiwu Huang and Luc Van Gool. A Riemannian Network for SPD Matrix Learning, In Proc. AAAI 2017.

Version 1.0, Copyright(c) November, 2017.

Note that the copyright of the manopt toolbox is reserved by https://www.manopt.org/

Usage

Step1: Place the used <a href="https://data.vision.ee.ethz.ch/zzhiwu/ManifoldNetData/SPDData/AFEW_SPD_data.zip"> AFEW </a> SPD data under the folder "./data/afew/". Note that the used <a href="https://data.vision.ee.ethz.ch/zzhiwu/ManifoldNetData/SPDData/HDM05_SPDData.zip"> HDM05 </a> and <a href="https://data.vision.ee.ethz.ch/zzhiwu/ManifoldNetData/SPDData/PaSC_SPDData.zip"> PaSC </a> SPD data are also publicly available.

Step2: Launch spdnet_afew.m for a simple example.

Related Work/Implementation

1. Thanks to Oleg Smirnov who is Sr. Applied Scientist at Amazon, a <a href="https://github.com/master/tensorflow-manopt"> TensorFlow ManOpt </a> library is released to reproduce our SPDNet.

2. A NeurIPS 2019 paper "Riemannian batch normalization for SPD neural networks" develops batch normalization layer upon our SPDNet, with the official <a href="https://proceedings.neurips.cc/paper/2019/hash/6e69ebbfad976d4637bb4b39de261bf7-Abstract.html"> PyTorch code </a> being publicly available at the 'Supplemental' tab.

3. A report <a href="https://core.ac.uk/download/pdf/231946513.pdf"> "Second-order networks in PyTorch" </a> is released.

4. Thanks to Alireza Davoudi, there is another <a href="https://github.com/adavoudi/spdnet"> Python implementation </a> for SPDNet.

5. A direct extension of our SPDNet for facial emotion recognition is published by CVPR workshop 2018, with the <a href="https://github.com/d-acharya/CovPoolFER"> code </a> being available.

How to Cite <a name="How-to-Cite"></a>

If you find this project helpful, please consider citing us as follows:

@inproceedings{huang2017spdnet,
      title = {A Riemannian Network for SPD Matrix Learning},
      author    = {Huang, Zhiwu and
                   Van Gool, Luc},
      year = {2017},
      booktitle = {Association for the Advancement of Artificial Intelligence (AAAI)}
}