WildAvatar
收藏资源简介:
WildAvatar数据集由华中科技大学、南洋理工大学、大湾区大学和上海人工智能实验室联合创建,是一个大规模的野外人类头像创建数据集。该数据集从YouTube上提取,包含超过10,000个不同的人类主体和场景,数据量丰富,至少是现有数据集的10倍。数据集创建过程中采用了高效的自动收集和注释流程,确保了数据的高质量。WildAvatar数据集主要应用于VR/AR、电影制作、元宇宙等领域的3D人类头像创建,旨在解决现有实验室数据集在真实世界应用中的局限性。
WildAvatar, a large-scale wild human avatar creation dataset, was jointly created by Huazhong University of Science and Technology, Nanyang Technological University, Guangdong-Hong Kong-Macao Greater Bay Area University, and Shanghai AI Laboratory. Extracted from YouTube, this dataset includes over 10,000 distinct human subjects and scenarios, with a scale at least 10 times that of existing datasets. An efficient automated collection and annotation pipeline was adopted during the dataset construction process to ensure the high quality of the data. WildAvatar is primarily applied to 3D human avatar creation in fields such as VR/AR, filmmaking, and the metaverse, aiming to address the limitations of current laboratory datasets in real-world applications.

- 1WildAvatar: Web-scale In-the-wild Video Dataset for 3D Avatar Creation华中科技大学, 南洋理工大学, 大湾区大学, 上海人工智能实验室 · 2024年



