Paper | Video | Project Page
We have tested our code on the following setups:
- Ubuntu 20.04 LTS
- Windows 10, 11
- Python >= 3.8
- Pytorch >= 1.11
- conda >= 4.9.2 (optional but recommended)
Follow these commands to create a conda environment:
conda create -n IDMS python=3.8
conda activate IDMS
conda install -c pytorch pytorch=1.11 torchvision cudatoolkit=11.3
pip install -r requirements.txt
For pytorch3D installation refer to https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md
Note: If PyOpenGL installed using requirements.txt
causes issues in Ubuntu, then install PyOpenGL using:
apt-get update
apt-get install python3-opengl
-
Follow the instructions on the SMPL-X website to download SMPL-X model and keep the downloaded files under the
smplx_model
folder. -
Download the GRAB dataset from the GRAB website, and follow the instructions there to extract the files. Save the raw data in
../DATASETS/GRAB
. -
To pre-process the GRAB dataset for our setting, run:
python src/data_loader/dataset_preprocess.py
Download the pretrained weights for the models used in our paper from here and keep it inside save\pretrained_models
.
- To evaluate our pre-trained model, run:
python src/evaluate/eval.py
- To generate the
.npy
files with the synthesized motions, run:
python src/test/test_synthesis.py
- To visualize sample results from our paper, run:
python src/visualize/render_smplx.py
-
To train our synthesis modules:
a. To train the Arm Synthesis module, run:
python src/train/train_arms.py
b. To train the Body Synthesis module, run:
python src/train/train_body.py
Keep the parameters the same as the pre-trained model argument file.
This code is distributed under MIT LICENSE.