If you find our work interesting and benifits your research, please consider citing:

@inproceedings{ren2022extrude,
	title = {ExtrudeNet: Unsupervised Inverse Sketch-and-Extrude for Shape Parsing},
	author = {Ren, Daxuan and Zheng, Jianmin and Cai, Jianfei and Li, Jiatong and Zhang, Junzhe},
	booktitle = {ECCV},
	year = {2022}}

We recommand using Anaconda to set the envoriment, once Anacodna in installed, run the following command.

conda create --name ExtrudeNet python=3.7
conda activate ExtrudeNet
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=10.1 -c pytorch
conda install -c open3d-admin open3d
conda install numpy
conda install pymcubes
conda install tensorboard
conda install scipy
pip install tqdm

As the processed dataset is extreamly large, we cannot provide a ready to download dataset. To prepare the dataset:

1. Download ShapeNet dataset (remember to agree their EULA)

   mkdir data
   cd data
   wget http://shapenet.cs.stanford.edu/shapenet/obj-zip/ShapeNetCore.v1.zip
   

2. unzip the desired category for example Plane (02691156)

   unzip ShapeNetCore.v1.zip
   cd ShapeNetCore.v1
   unzip 02691156.zip
   

3. Build "watertight" and add it to PATH

   cd ../../
   git clone https://github.com/skanti/generate-watertight-meshes-and-sdf-grids.git
   cd generate-watertight-meshes-and-sdf-grids
   mkdir build
   cd build
   cmake ..
   make -j
   export PATH=$PATH:`pwd`
   

4. Build triangle hash

   cd datasets
   python setup.py install
   

5. Edit the categories you want to run in the pre-processing script and run it

   vim preprocess.py
   python preprocess.py
   

python train.py --config_path ./configs/plane.json

python eval.py --config_path ./configs/plane.json

python metrics.py --config_path ./configs/plane.json

This project is licensed under the terms of the MIT license (see LICENSE for details).