Sign Language Segmentation Training

This repository holds the code to performing BIO tagging for sign language segmentation. This is a pre-processing step for tasks such as sign language transcription or sign language translation.

Setup

Install with pip, preferably inside a Python 3 virtual environment:

pip install git+https://github.com/bricksdont/sign-segmentation

This will install two entry points to the code, sign-language-segmentation-train and sign-language-segmentation-create-tfrecord.

Creating a tfrecord file

sign-language-segmentation-create-tfrecord -h

Training a model

sign-language-segmentation-train -h

Running tests

To run tests with coverage, execute

pytest

Citations

# Datasets library
@misc{moryossef2021datasets, 
    title={Sign Language Datasets},
    author={Moryossef, Amit},
    howpublished={\url{https://github.com/sign-language-processing/datasets}},
    year={2021}
}

# The Public DGS Corpus
@inproceedings{hanke2020extending,
  title={{E}xtending the {P}ublic {DGS} {C}orpus in Size and Depth},
  author={Hanke, Thomas and Schulder, Marc and Konrad, Reiner and Jahn, Elena},
  booktitle={Proceedings of the LREC2020 9th Workshop on the Representation and Processing of Sign Languages: Sign Language Resources in the Service of the Language Community, Technological Challenges and Application Perspectives},
  pages={75--82},
  year={2020}
}