This repository includes the rendering pipeline for SynthSL: Expressive Humans for Sign Language Image Synthesis (FG 2024), to render synthetic data based on the SMPL-x model.
SynthSL is publicly available for research purposes. To get access to the dataset, please send an email to jilliam_maria.diaz_barros [at] dfki [dot] de.
The implementation of our SwinGenerator and sign image synthesis architecture can be forked from here.
In progress...
If you find this repository useful for your research, please cite the following paper:
@inproceedings{diazbarros2024SynthSL,
title={SynthSL: Expressive Humans for Sign Language Image Synthesis},
author={D{\'\i}az Barros, Jilliam Mar{\'\i}a and Wang, Chen-Yu and Malik, Jameel and Arafa, Abdalla and Stricker, Didier},
booktitle={Int. Conf. on Automatic Face and Gesture Recognition (FG)},
year={2024},
organization={IEEE}
}
This repository is partially based on the code from SURREAL and ObMan.