zahemen9900/signtalk-gh-sampled
收藏Hugging Face2025-12-12 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/zahemen9900/signtalk-gh-sampled
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是通过合并SignTalk-GH语料库和GSL字典（zenodo openpose视频）并采样到一个平衡子集而生成的，旨在用于下游预处理和Text2Sign管道。数据集包含视频文件和元数据文件，元数据包括视频文件名称、句子ID、句子文本、类别和变体等信息。数据集还提供了详细的创建步骤、加载方法、推荐的分割方式、许可信息、已知的注意事项以及来源和可重复性说明。
Dataset generated by merging the SignTalk-GH corpus with the GSL dictionary (zenodo openpose videos) and sampling to a balanced subset. Intended for downstream preprocessing and Text2Sign pipelines. The dataset includes video files and metadata files, with metadata containing video file names, sentence IDs, sentence text, categories, and variants. The README also provides detailed creation steps, loading methods, recommended splits, licensing information, known caveats, and provenance & reproducibility details.
提供机构：
zahemen9900
