Unsupervised Learning of Robust Spectral Shape Matching (SIGGRAPH/TOG 2023)

Installation

conda create -n fmnet python=3.8 # create new viertual environment
conda activate fmnet
conda install pytorch cudatoolkit -c pytorch # install pytorch
pip install -r requirements.txt # install other necessary libraries via pip

Dataset

To train and test datasets used in this paper, please download the datasets from the this link and put all datasets under ../data/

├── data
    ├── FAUST_r
    ├── FAUST_a
    ├── SCAPE_r
    ├── SCAPE_a
    ├── SHREC19_r
    ├── TOPKIDS
    ├── SMAL_r
    ├── DT4D_r
    ├── SHREC20
    ├── SHREC16
    ├── SHREC16_test

We thank the original dataset providers for their contributions to the shape analysis community, and that all credits should go to the original authors.

Data preparation

For data preprocessing, we provide preprocess.py to compute all things we need. Here is an example for FAUST_r.

python preprocess.py --data_root ../data/FAUST_r/ --no_normalize --n_eig 200

Train

To train the model on a specified dataset.

python train.py --opt options/train/faust.yaml 

You can visualize the training process in tensorboard.

tensorboard --logdir experiments/

Test

To test the model on a specified dataset.

python test.py --opt options/test/faust.yaml 

The qualitative and quantitative results will be saved in results folder.

Texture Transfer

An example of texture transfer is provided in texture_transfer.py

python texture_transfer.py

Pretrained models

You can find all pre-trained models in checkpoints for reproducibility.

Partial Shape Matching on SHREC’16

There were two issues with the partial shape matching experiments on the SHREC'16 dataset related to the training/test splits [Bracha et al. 2023, Ehm et al. 2024]. Below, we provide additional evaluations that substantiate our claims:

Geo err (x100)	CUTS on CUTS	CUTS on HOLES	HOLES on CUTS	HOLES on HOLES	CUTS on CUTS'24*	HOLES on CUTS'24*
Ours original**	3.3	13.7	5.2	9.1	3.4	5.5
Ours new***	3.2	13.5	5.6	8.2	3.2	5.9

* CUTS'24 refers to the new test split from [Ehm et al. 2024], the split can be found here.

** Pretrained on TOSCA

*** Pretrained on FAUST + SCAPE + SMAL + DT4D-H, 25 test-time adaptation iterations

[Bracha et al. 2023] A. Bracha, T. Dages, R. Kimmel, On Partial Shape Correspondence and Functional Maps, arXiv 2023.

[Ehm et al. 2024] V. Ehm, M. Gao, P. Roetzer, M. Eisenberger, D. Cremers, F. Bernard, Partial-to-Partial Shape Matching with Geometric Consistency, CVPR 2024.

Acknowledgement

The implementation of DiffusionNet is based on the official implementation.

The framework implementation is adapted from Unsupervised Deep Multi Shape Matching.