On a Linear Gromov–Wasserstein Distance

This repository contains the code for the paper 'On a Linear Gromov–Wasserstein Distance'. The paper has been published in IEEE TIP. A preprint version is available on arXiv.

Please cite the paper if you use the code.

Citation

1. Florian Beier, Robert Beinert, Gabriele Steidl, 'On a Linear Gromov-Wasserstein Distance', IEEE Transactions on Image Processing, 31, 7292-7305, 2022.

Experiments

The numerical simulations can be reproduced using the script

• ellipses.ipynb for the experiments in 'Gromov-Wasserstein of Elliptical Disks' (Figs 1-3; Tab I),

• 2d_shapes.ipynb for the experiments in 'Gromov-Wasserstein in 2D Shape Analysis' (Figs 4-7; Tab II),

• 3d_shapes_2.ipynb for the experiments in 'Gromov-Wasserstein in 3D Shape Analysis' (Fig 8-10; Tabs III-V),

Some of the input data is based on The 2D Shape Structure Dataset and Mesh Data from Deformation Transfer for Triangle Meshes.

Contributing

The code is available under a MIT license.