Spectral Meets Spatial: Harmonising 3D Shape Matching and Interpolation

Installation

conda create -n sms python=3.9 # create new viertual environment
conda activate sms
conda install pytorch=1.13.0 torchvision pytorch-cuda=11.6 -c pytorch -c nvidia # install pytorch
conda install -c fvcore -c iopath -c conda-forge fvcore iopath
conda install pytorch3d -c pytorch3d # install pytorch3d
conda install pyg -c pyg # install torch_geometric
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_aligned

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_aligned/ --no_normalize --n_eig 300

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.

Shape interpolation

To perform shape interpolation on a specified dataset.

python shape_interpolation.py

The results will be saved in results folder.

Acknowledgement

The implementation of DiffusionNet is based on the official implementation.

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