Abstract
We release six datasets aimed at sign language translation research. Raw data was collected by broadcast partners SWISSTXT and VRT between the March-Aug 2020 period. Manual subtitle-sign video alignment has been done on a subset of the data and released for research purposes. Dataset contains anonymized sign language videos, aligned and raw subtitles and 2D/3D human skeletal pose information.
For further information and to access the dataset, please visit: https://www.cvssp.org/data/c4a-news-corpus/