a new multi-view dataset for human-object interactions
收藏资源简介:
该数据集由Saarland University和Max Planck Institute for Informatics共同创建,包含多个场景中人类与不同大小物体的自然交互。数据集通过大型的捕捉穹顶拍摄,旨在为3D重建提供具有挑战性的交互场景,特别是在物体间存在严重遮挡和边界模糊的情况下。数据集的应用领域是3D重建,主要解决的是在多视角RGB图像中分离多个物体的几何形状的问题。
This dataset was co-created by Saarland University and the Max Planck Institute for Informatics. It contains natural human-object interactions of varying scales across multiple scenes, captured using a large-scale capture dome. The dataset is designed to provide challenging interactive scenarios for 3D reconstruction, particularly in cases with severe inter-object occlusions and ambiguous object boundaries. Its application domain focuses on 3D reconstruction, where it primarily addresses the problem of separating the geometric shapes of multiple objects from multi-view RGB images.




