cloth_shape_estimation

Dependencies

1) DiffCloth

Clone the repo:

git clone --recursive https://github.com/friolero/DiffCloth.git

Build the Python binding:

cd DiffCloth
python setup.py install --user

2) Pytorch3D

The complete official installation can be refered here. This is the installation without using Conda.

Install cuda 11.6

wget https://developer.download.nvidia.com/compute/cuda/11.6.0/local_installers/cuda_11.6.0_510.39.01_linux.run
sudo chmod +x cuda_11.6.0_510.39.01_linux.run
sudo ./cuda_11.6.0_510.39.01_linux.run

Install Nvidia cub

curl -LO https://github.com/NVIDIA/cub/archive/1.10.0.tar.gz
tar xzf 1.10.0.tar.gz
export CUB_HOME=$PWD/cub-1.10.0

Install dependencies and Pytorch3d

pip install -U fvcore iopath 
pip install scikit-image matplotlib imageio plotly opencv-python
pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu116
pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu116_pyt1130/download.html

Usage

1. Generate perturbed cloth data as .obj files

python3 cloth_randomization.py -mode 1 -output_dir cloth_project -n_output 5000 -seed SEED -n_openmp_thread N_PARALLEL_THREAD

Wavefront Data will be saved under $DIFFCLOTH_PATH/output/$args.output_dir.

• Mode 0 for randomized perturbation trajectory;
• Mode 1 uses Bezier curve to generate the perturbation path from a randomly identified control vertex to another vertex position within a specified distance range.

2. Data driven deformation estimation with deform-net

Create rendered image from perturbed cloth obj files and partition data into training, evaluation and testing set. This may takes 1-2 hours. python3 data_utils.py

Start deform_net training with: python3 deform_net_train.py

Perform inference with trained deform_net : python3 deform_net_predict.py -m model.ckpt -obj PATH_TO_OBJ

3. Cloth deformation optimization via differentiable rendering

python3 cloth_deform_estimation.py

• Step (2) is optional to run step (3)