In this paper, we address the problem of wide-baseline camera pose estimation from a group of 360∘ panoramas under upright-camera assumption. Recent work has demonstrated the merit of deep-learning for end-to-end direct relative pose regression in 360∘ panorama pairs [11]. To exploit the benefits of multi-view logic in a learning-based framework, we introduce Graph-CoVis, which non-trivially extends CoVisPose [11] from relative two-view to global multi-view spherical camera pose estimation. Graph-CoVis is a novel Graph Neural Network based architecture that jointly learns the co-visible structure and global motion in an end-to-end and fully-supervised approach. Using the ZInD [4] dataset, which features real homes presenting wide-baselines, occlusion, and limited visual overlap, we show that our model performs competitively to state-of-the-art approaches.
@article{arxiv.2304.13201,
title = {Graph-CoVis: GNN-based Multi-view Panorama Global Pose Estimation},
author = {Negar Nejatishahidin and Will Hutchcroft and Manjunath Narayana and Ivaylo Boyadzhiev and Yuguang Li and Naji Khosravan and Jana Kosecka and Sing Bing Kang},
journal= {arXiv preprint arXiv:2304.13201},
year = {2023}
}