Common Objects in 3D (CO3Dv2)

该数据集包含了带有相机姿态标注的真实世界物体，用于评估从稀疏视图中进行新颖视图合成的技术。该数据集还包括了少量的视图训练集和开发集序列，这些序列被用于训练和评估任务，即从稀疏视图中进行新颖视图合成。

This dataset comprises real-world objects annotated with camera poses, and it is intended to evaluate techniques for novel view synthesis from sparse views. It also incorporates a small number of view sequences for the training and development sets, which are utilized to train and evaluate the task of novel view synthesis from sparse views.