NeRFDeformer Dataset

该数据集包含了113个合成场景，这些场景利用了Objaverse数据集中的47个动态对象模型，特别关注于复杂的非刚性变换。此外，该数据集还包括与渲染图像对应的深度图，用于评估基于单次变换场景观察的NeRF编辑方法和生成模型的有效性。规模上，它涵盖了47个动态对象模型中的113个场景，其任务在于基于单次观察修改经过变换场景的NeRF表示。

This dataset comprises 113 synthetic scenes constructed using 47 dynamic object models sourced from the Objaverse dataset, with a particular emphasis on complex non-rigid transformations. Additionally, the dataset includes depth maps paired with the rendered images, which are used to evaluate the effectiveness of NeRF editing methods and generative models based on single-view observations of transformed scenes. In terms of scale, this dataset covers 113 scenes derived from the 47 dynamic object models, and its core task is to modify the NeRF representation of a transformed scene based on a single-view observation.