We introduce Princeton365, a diverse dataset of 365 videos with accurate camera poses, captured using a novel framework combining 360° cameras and calibration boards. It enables robust SLAM and novel view synthesis evaluation across challenging indoor, outdoor, and object-centric scenes.
Stamatis Alexandropoulos*,
Karhan Kayan*,
Rishabh Jain,
Yiming Zuo,
Erich Liang,
Jia Deng
* Equal contribution (random order)
Arxiv
This work was partially supported by the National Science Foundation, Onassis Foundation and a gift from Meta Reality Labs.
