DNA-Rendering

DNA-Rendering 数据集涵盖多人种，多年龄段，多民族，多朝代及多种特殊技能的人体数据特性，是目前对人体形态、动作、衣物等细节多视角、高分辨采集中规模最大的数据集，以多样性、高保真、动态性为优势。 √密集视角采集：60高清视角，8个Kinect深度图视角 √4K超高清动作捕捉 √表演场景丰富：日常动作，专业场景 √专业技能动作采集 √海量现代、古典及民族服饰 √大量人物交互场景 √多元数据及标注：图片、分割、点云、人体关键点及模型The DNA-Rendering dataset contains human body data characteristics spanning multiple races, age groups, ethnicities, historical dynasties, and various special skills. It is currently the largest-scale dataset for multi-view, high-resolution capture of details such as human morphology, movements, and clothing, with prominent advantages in diversity, high fidelity, and dynamic performance. - Dense multi-view data acquisition: 60 high-definition viewpoints, 8 Kinect depth map viewpoints - 4K ultra-high-definition motion capture - Rich performance scenarios including daily movements and professional settings - Acquisition of professional skill-based movements - Vast library of modern, classical, and ethnic garments - Abundant human-human interaction scenarios - Diverse data types and annotations: images, segmentation masks, point clouds, human body keypoints, and 3D human models