PRAGO: Differentiable Multi-View Pose Optimization From Objectness Detections*

Taiana, Matteo; Toso, Matteo; James, Stuart; Bue, Alessio Del

Abstract

Robustly estimating camera poses from a set of images is a fundamental task which remains challenging for differentiable methods, especially in the case of small and sparse camera pose graphs. To overcome this challenge, we propose Pose-refined Rotation Averaging Graph Optimization (PRAGO). From a set of objectness detections on unordered images, our method reconstructs the rotational pose, and in turn, the absolute pose, in a differentiable manner benefiting from the optimization of a sequence of geometrical tasks. We show how our objectness pose-refinement module in PRAGO is able to refine the inherent ambiguities in pairwise relative pose estimation without removing edges and avoiding making early decisions on the viability of graph edges. PRAGO then refines the absolute rotations through iterative graph construction, reweighting the graph edges to compute the final rotational pose, which can be converted into absolute poses using translation averaging. We show that PRAGO is able to outperform non-differentiable solvers on small and sparse scenes extracted from 7-Scenes achieving a relative improvement of 21% for rotations while achieving similar translation estimates.

Citation

Taiana, M., Toso, M., James, S., & Bue, A. D. (2024, March). PRAGO: Differentiable Multi-View Pose Optimization From Objectness Detections*. Presented at 2024 International Conference on 3D Vision (3DV), Davos, Switzerland

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2024 International Conference on 3D Vision (3DV)
• Start Date: Mar 18, 2024
• End Date: Mar 21, 2024
• Acceptance Date: Oct 18, 2023
• Online Publication Date: Jun 12, 2024
• Publication Date: Jun 12, 2024
• Deposit Date: Sep 27, 2024
• Publicly Available Date: Oct 1, 2024
• Publisher: Institute of Electrical and Electronics Engineers
• Peer Reviewed: Peer Reviewed
• Volume: 2
• Pages: 324-333
• Series ISSN: 2378-3826
• Book Title: 2024 International Conference on 3D Vision (3DV)
• ISBN: 9798350362466
• DOI: https://doi.org/10.1109/3dv62453.2024.00117
• Public URL: https://durham-repository.worktribe.com/output/2879998