MVP-Human
收藏资源简介:
MVP-Human数据集是由中国科学院自动化研究所创建的大型3D穿着人体数据集,旨在解决从多帧图像中重建3D穿着人体模型的问题。该数据集包含400个不同身份的个体,每个个体有15个不同姿势的扫描和每姿势8视角的图像,总计6000个3D扫描和48000张图像。数据集的创建过程涉及复杂的捕捉设置和高质量的3D扫描技术,确保了数据的多样性和准确性。MVP-Human数据集的应用领域广泛,包括虚拟现实/增强现实体验、视频编辑和虚拟试衣等,旨在推动3D人体重建技术的发展。
MVP-Human Dataset is a large-scale 3D clothed human dataset developed by the Institute of Automation, Chinese Academy of Sciences, aiming to address the challenge of reconstructing 3D clothed human models from multi-frame images. This dataset includes 400 distinct individuals, each with scans captured in 15 different poses and images taken from 8 viewpoints per pose, totaling 6000 3D scans and 48,000 images. The construction of the MVP-Human Dataset involves sophisticated capture setups and high-quality 3D scanning technologies, ensuring the diversity and accuracy of the data. The MVP-Human Dataset has a wide range of application scenarios, including virtual reality (VR) and augmented reality (AR) experiences, video editing, virtual try-on, and more, and is designed to promote the advancement of 3D human reconstruction technologies.




