研究人员推出了 SignVerse-2M，这是一个包含两百万个视频片段的大规模数据集，涵盖25种以上的手语。该数据集是基于姿态的原生数据，意味着它使用DWPose将原始视频转换为2D姿态序列。这种方法旨在通过关注姿态而非外观来提高手语识别和生成模型的鲁棒性，使其更适合现实世界、开放场景的应用。 AI
影响 提供了一个新的、基于姿态的原生数据集，以推进手语识别和生成模型的研究。
排序理由 介绍用于手语研究的大规模数据集的新学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
Existing large-scale sign language resources typically provide supervision only at the level of raw video-text alignment and are often produced in laboratory settings. While such resources are important for semantic understanding, they do not directly provide a unified interface …
arXiv:2605.01720v1 Announce Type: new Abstract: Existing large-scale sign language resources typically provide supervision only at the level of raw video-text alignment and are often produced in laboratory settings. While such resources are important for semantic understanding, t…