Sihang Li* · Siqi Tan* · Bowen Chang · Jing Zhang · Chen Feng · Yiming Li
* Equal contribution
TLDR: We make camera localization more generalizable by addressing the data gap via 3DGS and learning gap via a two-branch joint learning with adversarial loss, achieving localization accuracy surpassing 1cm/0.3° in indoor scenarios, 20cm/0.5° in outdoor scenarios, and 10cm/0.2° in driving scenarios.
- 2024/12/1: Our paper is now available on arXiv!
The code for appearance-varying 3DGS, RAP, and RAPref will be released soon!
This work was supported in part through NSF grants 2238968, 2121391, and 2024882, and the NYU IT High Performance Computing resources, services, and staff expertise. Yiming Li is supported by NVIDIA Graduate Fellowship.
If you find our work helpful, please consider citing our paper!
@article{Li2024unleashing,
title={Unleashing the Power of Data Synthesis},
author={Sihang Li and Siqi Tan and Bowen Chang and Jing Zhang and Chen Feng and Yiming Li},
year={2024},
journal={arXiv preprint arXiv:2412.00138},
}