Home Original page

GitHub - Ruiyuan-Zhang/CCS: This repository contains the source codes for the paper "Scalable Geometric Fracture Assembly via Co-creation Space among Assemblers". Accepted by AAAI 2024, focusing on multi-parts issue of assembly.

Co-creation Space - Code Release

I. Introduction

We are excited to announce that our paper, titled "Scalable Geometric Fracture Assembly via Co-creation Space among Assemblers", has been accepted by the AAAI 2024. In line with our commitment to open science and reproducible research, we are preparing to release the code associated with our findings. Co-creation Space(CSS) is based on Multi-Part Shape Assembly.

If you encounter any issues while reproducing my code, feel free to contact me. I am always happy to offer assistance, ensuring that you can smoothly understand. Let's work together and progress, building a friendship along the way!

II. Traing codes

2.1 Data Preparation

Please visit link for instructions on how to download these datasets.

2.2 Virtual Environment Installation

conda create -n ccs python=3.8.5
conda activate ccs
conda install pytorch=1.10 torchvision torchaudio cudatoolkit=11.3 -c pytorch
conda install pytorch-lightning=1.6.2  # pip install lightning
conda install -c fvcore -c iopath -c conda-forge fvcore iopath
pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/download.html # conda install pytorch3d -c pytorch3d  

pip install setuptools==59.5.0 
pip install setproctitle
pip install trimesh
pip install wandb
pip install scipy
pip install pyntcloud
pip install matplotlib
pip install ninja

2.3 Module Installation

  1. Run pip install -e . to install this module
  2. Go to multi_part_assembly/utils/chamfer and run pip install -e .
  3. Go to multi_part_assembly/models/modules/encoder/pointnet2/pointnet2_ops_lib and run pip install -e .

2.4 Train Scripts

python scripts/train.py --project_name debug --gpu 0 --cfg configs/wx_transformer/wx_transformer/topk/partnet/FF-top10-partnet_chair.py

III. Test Codes

3.1 Test Scripts

# categories = ['BeerBottle', 'Bowl', 'Cup', 'DrinkingUtensil', 'Mug', 'Plate', 'Spoon', 'Teacup', 'ToyFigure', 'WineBottle', 'Bottle', 'Cookie', 'DrinkBottle', 'Mirror', 'PillBottle', 'Ring', 'Statue', 'Teapot', 'Vase', 'WineGlass']
python scripts/test.py --gpus 0 --category BeerBottle --cfg_file configs/wx_transformer/wx_transformer/topk/partnet/FF-top10-partnet_chair.py --weight checkpoint/FF-top10-partnet_chair/models/model-epoch=499.ckpt --min_num_part 1 --max_num_part 20

V. Stay Updated

To stay updated on our code release and other project news, please:

VI. License

This project is released under the MIT license.

VII. Acknowledgements

We would like to thank the AAAI reviewers and our colleagues who have contributed to this research. We are looking forward to sharing our work with the community and hope it will be a valuable resource. We also thank the benchmark provided by Ziyi Wu.