Spatio-temporal Sign Language Representation and Translation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918165564358656 |
|---|---|
| author | Hamidullah, Yasser van Genabith, Josef España-Bonet, Cristina |
| author_facet | Hamidullah, Yasser van Genabith, Josef España-Bonet, Cristina |
| contents | This paper describes the DFKI-MLT submission to the WMT-SLT 2022 sign language translation (SLT) task from Swiss German Sign Language (video) into German (text). State-of-the-art techniques for SLT use a generic seq2seq architecture with customized input embeddings. Instead of word embeddings as used in textual machine translation, SLT systems use features extracted from video frames. Standard approaches often do not benefit from temporal features. In our participation, we present a system that learns spatio-temporal feature representations and translation in a single model, resulting in a real end-to-end architecture expected to better generalize to new data sets. Our best system achieved $5\pm1$ BLEU points on the development set, but the performance on the test dropped to $0.11\pm0.06$ BLEU points. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_19413 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Spatio-temporal Sign Language Representation and Translation Hamidullah, Yasser van Genabith, Josef España-Bonet, Cristina Computation and Language Computer Vision and Pattern Recognition This paper describes the DFKI-MLT submission to the WMT-SLT 2022 sign language translation (SLT) task from Swiss German Sign Language (video) into German (text). State-of-the-art techniques for SLT use a generic seq2seq architecture with customized input embeddings. Instead of word embeddings as used in textual machine translation, SLT systems use features extracted from video frames. Standard approaches often do not benefit from temporal features. In our participation, we present a system that learns spatio-temporal feature representations and translation in a single model, resulting in a real end-to-end architecture expected to better generalize to new data sets. Our best system achieved $5\pm1$ BLEU points on the development set, but the performance on the test dropped to $0.11\pm0.06$ BLEU points. |
| title | Spatio-temporal Sign Language Representation and Translation |
| topic | Computation and Language Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.19413 |