Installation

Install Hand Embodiment

This software is implemented in Python. If you do not have a working Python environment so far, we recommend to install it with Anaconda. Then you can follow these instructions:

Clone repository:

git clone https://github.com/dfki-ric/hand_embodiment.git

Prepare MANO model:

  1. Download MANO model from here and unzip it to hand_embodiment/hand_embodiment/model/mano.

  2. Go to hand_embodiment/hand_embodiment/model/mano.

cd hand_embodiment/hand_embodiment/model/mano
  1. For this preparation step, please install following dependencies.

pip install numpy scipy chumpy
  1. Run prepare_mano.py, you will get the converted MANO model that is compatible with this project.

Warning

The downloaded pickle files of the MANO model uses np.bool, which is removed since numpy 1.24. Running prepare_mano.py with a numpy version >= 1.24 will result in an error. Please install a lower version of numpy, e.g. 1.23.5 or refer to the numpy version listed in requirements.txt, where the exact versions of the depedencies are listed, in the main folder. Alternatively, contact one of the maintainer to receive the converted MANO model under the licensing terms of the MANO publishers.

python prepare_mano.py  # convert model to JSON
  1. Go back to the main folder.

cd ../../../

Install hand_embodiment from main directory:

pip install -e .

The Python version used to produce the results in the paper was 3.8. Exact versions of the dependencies are listed in requirements.txt (pip install -r requirements.txt). Alternatively, you can create a conda environment from environment.yml:

conda env create -f environment.yml
conda activate hand_embodiment
pip install -e .
pytest test/  # verify installation

Data

The dataset is available at Zenodo:

https://zenodo.org/badge/DOI/10.5281/zenodo.7130208.svg

We recommend to use zenodo_get to download the data:

pip install zenodo-get
mkdir -p some/folder/in/which/the/data/is/located
cd some/folder/in/which/the/data/is/located
zenodo_get 7130208