jackzhousz / unsupervised-learning-of-robust-spectral-shape-matching

SIGGRAPH/TOG, 2023: Unsupervised Learning of Robust Spectral Shape Matching

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 the pre-trained models on SURREAL-5k dataset in checkpoints for reproducibility.

Acknowledgement

The implementation of DiffusionNet is based on the official implementation.

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