Introducing A Bangla Sentence - Gloss Pair Dataset for Bangla Sign Language Translation and Research

The launch of the Bangla-SGP dataset represents a pivotal moment for Bangla Sign Language (BdSL) translation, which has long been hindered by a lack of comprehensive datasets for sentence-level translation. With only word and alphabet level detection previously explored, this new dataset introduces 1,000 meticulously annotated sentence-gloss pairs, significantly improving the resources available for researchers and practitioners. The augmentation process, which adds around 3,000 synthetically generated pairs, employs advanced linguistic strategies and prompt engineering techniques, ensuring a robust dataset that enhances translation quality. This initiative not only addresses the pressing need for resources in low-resource NLP tasks but also sets a precedent for future research and development in sign language translation, ultimately fostering better communication and accessibility for the deaf community.