Dataset Hub

The dataset contains 3D models (in .glb format) and rendering images of buildings, sourced from the Sketchfab website. It supports both Chinese and English languages, with tags including architecture, 3D, and Sketchfab. The dataset size ranges from 100 B to 1 T.

The NavVis Indoor Dataset is a collection of over 50,000 high‑resolution geo‑referenced images, covering more than 50,000 m² of indoor space across twelve different buildings at the Technical University of Munich. The dataset includes external poses for all images, captured between August 2015 and March 2016, encompassing various indoor environments such as different architectural styles and lighting conditions.

The dataset collects RGB images, triangulated meshes with UV coordinates, texture maps, 3D human poses, and camera projection matrices from the NBA2K19 game engine. It features diverse basketball postures and provides industry‑grade meshes of clothed basketball players.