Deep Fashion3D
收藏资源简介:
Deep Fashion3D是由香港中文大学(深圳)创建的大规模3D服装模型数据集,包含2078个从真实服装重建的模型,覆盖10个不同类别和563个服装实例。该数据集提供丰富的标注,包括3D特征线、3D身体姿势和对应的多视角真实图像。每个服装模型都随机摆放在假人模型或真人上,以增强真实服装变形的多样性。Deep Fashion3D旨在为基于图像的服装重建系统建立新的基准和数据集,解决3D服装建模的难题,特别是在缺乏大规模服装模型的情况下。
Deep Fashion3D is a large-scale 3D clothing model dataset developed by The Chinese University of Hong Kong, Shenzhen. It contains 2078 models reconstructed from real clothing, covering 10 distinct categories and 563 clothing instances. This dataset provides rich annotations including 3D feature lines, 3D body poses, and corresponding multi-view real images. Each clothing model is randomly placed on a mannequin or a real human body to enhance the diversity of realistic clothing deformations. Deep Fashion3D aims to establish new benchmarks and serve as a dedicated dataset for image-based clothing reconstruction systems, addressing the challenges in 3D clothing modeling, especially when large-scale clothing models are scarce.




