Project | Paper | Video

Our environment dependencies are included in the "BreakingBad\multi_part_assembly.egg-info\requires.txt" file. You can easily install them when you create a new conda environment.

For Breaking Bad dataset, please download the dataset following https://breaking-bad-dataset.github.io/.

For Geometric Shape Mating dataset, please generate the dataset following https://github.com/pairlab/NSM. Besides, you can download the dataset we generate in https://www.dropbox.com/scl/fi/xigal4s4xmksie4ihm7su/ShapeNet_0103.zip?rlkey=54f9w1fcnc2fjm5xmijpmkzf5&dl=0.

To train models for Breaking Bad dataset, pleas run

cd BreakingBad
python scripts/train.py --cfg_file configs/vnn/vnn-everyday.py

To train models for Geometric Shape Mating dataset, pleas run

cd NSM
python script/train_eqv_CR.py --cfg_file train_vnn_pn.yml

The evaluation of models in Breaking Bad dataset is automatically performed during training.

To evaluate models for Geometric Shape Mating dataset, pleas run

cd NSM
python script/eval_eqv_CR.py --cfg_file eval_vnn_pn.yml

If you find this paper useful, please consider citing:

@InProceedings{Wu_2023_ICCV,
    author    = {Wu, Ruihai and Tie, Chenrui and Du, Yushi and Zhao, Yan and Dong, Hao},
    title     = {Leveraging SE(3) Equivariance for Learning 3D Geometric Shape Assembly},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2023},
    pages     = {14311-14320}
}

If you have any questions, please feel free to contact Ruihai Wu at wuruihai_at_pku_edu_cn and Chenrui Tie at crtie_at_pku_edu_cn