高精人体三维数据库

高保真“人体-站姿”三维扫描数据集，以真实采集的静态3D扫描样本为主，用于研究三维人体几何重建、人体姿态估计、3D资产生成，动作生成等任务。 数据集规模与组成如下： 样本数量：5182组三维交互扫描（每个样本为一次独立扫描）。 人体被试：包含不同性别、身材、服饰，动作等。 主要数据项（按样本提供）： 纹理化三维网格模型（Wavefont OBJ, 含材质文件MTL与纹理贴图PNG）。 实际提交数据量：762GBHigh-fidelity "Human-Standing Pose" 3D Scanning Dataset is mainly composed of real-collected static 3D scanning samples, which is targeted for research tasks including 3D human geometric reconstruction, human pose estimation, 3D asset generation, motion generation and other related fields. The scale and composition of the dataset are as follows: Number of samples: 5182 sets of 3D interactive scans, where each sample corresponds to an independent scanning session. Human subjects: Cover diverse genders, body types, clothing styles and postures. Main data items (provided per sample): Textured 3D mesh models in Wavefront OBJ format, with accompanying MTL material files and PNG texture maps. Total actual submitted data volume: 762 GB

搜集汇总

背景与挑战

背景概述

该数据集是一个高保真的人体三维扫描数据库，包含5182组静态三维交互扫描样本，覆盖不同性别、身材和服饰。它主要用于三维人体几何重建、姿态估计和3D资产生成等研究任务，总数据量为762GB。