Installation¶
Requirements¶
- Python >= 3.10
Install from PyPI¶
pip install sign-language-tools
This installs the core dependencies: matplotlib, mediapipe, numpy, opencv-python, pandas,
plotly, scipy, scikit-learn, torch, tqdm and vidgear.
Note
The VideoPlayer's random-access video backend prefers PyAV
or decord when available, falling back to opencv-python
otherwise. Install one of them for faster seeking on long videos:
pip install "sign-language-tools[video-backends]"
Install for development¶
Clone the repository and install it in editable mode with the dev extra, which adds testing and
documentation tooling (pytest, mkdocs, mkdocs-material, mkdocstrings):
git clone https://github.com/ppoitier/sign-language-tools.git
cd sign-language-tools
pip install -e ".[dev]"
Serve the documentation locally with:
mkdocs serve
Pairing with sign-language-data-loading¶
Most examples in this documentation load sample data with
sign-language-data-loading, a separate
package for loading sign language datasets stored as WebDataset shards:
pip install sign-language-data-loading
See Working with sign-language-data-loading for details.