PG-3DGS Supporting Dataset: Multiview Teapot and Airplane Images, Depth Maps, Masks, and Camera Parameters

This dataset contains the processed experimental inputs used in the PG-3DGS study. It includes 19 object instances in total: 15 teapots and 4 airplanes. For each instance, the dataset provides 20 views with RGB images, depth maps, object masks, and corresponding camera parameters/angles. These files are intended to support reproducibility of the multiview reconstruction and physics-guided optimization experiments reported in the paper. The dataset is organized by object, with per-view image, depth, mask, and camera metadata files.

本数据集包含PG-3DGS研究中使用的经预处理的实验输入数据。数据集共包含19个物体实例：15个茶壶与4架飞机。针对每个实例，数据集提供20组视图数据，每组包含RGB图像、深度图、物体掩码以及对应的相机参数与拍摄角度。这些文件旨在支撑论文中报道的多视图重建与物理引导优化实验的可复现性。本数据集按物体进行分组组织，包含每视图对应的图像、深度图、掩码文件与相机元数据文件。