2026-10-09T22:16:50Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33933409 2026-10-09T13:26:57Z category_58319 portal_201 item_type_3 month_year_10_2026

NGT-LEX-UvA-2025-video_glos_raw_part27 Gomèr Otterspeer (22384924) Language, Communication and Culture sign language Dutch Sign Language NGT Signbank sign gloss non-manual markers video corpus pose estimation <p>This dataset is part of the data collection NGT-LEX-UvA-2025-video (<a href="https://doi.org/10.21942/uva.c.8723623">https://doi.org/10.21942/uva.c.8723623</a>).</p><h3>About the content of the data collection</h3><p>The NGT-LEX-UvA-2025-video data collection contains video recordings of lexical signs in Dutch Sign Language (NGT). Each sign has been recorded with five cameras, simultaneously, from different angles (M/L/R/A/B).</p><p>Datasets with ‘nmm’ in their name contain video recordings of signs including non-manual markers (e.g. facial expressions and mouth movements). Datasets with ‘gloss’ in their name contain video recordings of signs that are purposely performed without non-manual markers, even though when used in context these signs would typically be performed with non-manual markers.</p><p>The collection also contains three types of data derived from the original video data, namely MediaPipe pose estimation data, SAM 3D body keypoint estimation data and HaMeR handshape estimation data.</p><p>See the NGT-LEX-UvA-2025-video_codebook dataset, which is part of this collection, for more details on the structure of the collection, file naming conventions, and a file mapping table.</p><h3>About the creation of the data collection</h3><p>All videos have been recorded at the SignLab video recording studio at the University of Amsterdam in 2023-2025. The data collection methods are described in detail in the following paper:<br><a href="https://aclanthology.org/2024.signlang-1.19/">https://aclanthology.org/2024.signlang-1.19/</a></p><h3>License</h3><p>All data is shared under a CC-BY 4.0 license.</p><h3>Funding</h3><p>The creation of this data collection has been supported by the Platform Digital Infrastructure for the Social Sciences and the Humanities (PDI-SSH) in the Netherlands.</p><h3>Accessibility</h3><p>The video data is also accessible through the Global Signbank platform, where each video is associated with semantic and phonological information:<br><a href="https://signbank.cls.ru.nl/datasets/NGT/">https://signbank.cls.ru.nl/datasets/NGT/</a></p><h3>Related data collection</h3><p>The related data collection NGT-LEX-UvA-2025-animations (<a href="https://doi.org/10.21942/uva.c.8733523">https://doi.org/10.21942/uva.c.8733523</a>) provides 3D animations, created with motion capture equipment, for 1,736 lexical signs in the current data collection.</p><h3>About this dataset</h3><p>NGT-LEX-UvA-2025-video NGT sign gloss (lexical sign) recordings: raw (unprocessed) multi-camera videos (M/L/R/A/B) (part 27, in recording-date order). See NGT-LEX-UvA-2025-video_codebook for file naming and file_mapping.csv.</p> 2026-10-09T13:26:57Z Dataset Dataset 10.21942/uva.33933409.v2 https://figshare.com/articles/dataset/Signbank_glos_raw_part27/33933409 CC BY 4.0