RESfM: Robust Deep Equivariant Structure from Motion

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Khatib, Fadi, Kasten, Yoni, Moran, Dror, Galun, Meirav, Basri, Ronen
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866913998784430080
author Khatib, Fadi
Kasten, Yoni
Moran, Dror
Galun, Meirav
Basri, Ronen
author_facet Khatib, Fadi
Kasten, Yoni
Moran, Dror
Galun, Meirav
Basri, Ronen
contents Multiview Structure from Motion is a fundamental and challenging computer vision problem. A recent deep-based approach utilized matrix equivariant architectures for simultaneous recovery of camera pose and 3D scene structure from large image collections. That work, however, made the unrealistic assumption that the point tracks given as input are almost clean of outliers. Here, we propose an architecture suited to dealing with outliers by adding a multiview inlier/outlier classification module that respects the model equivariance and by utilizing a robust bundle adjustment step. Experiments demonstrate that our method can be applied successfully in realistic settings that include large image collections and point tracks extracted with common heuristics that include many outliers, achieving state-of-the-art accuracies in almost all runs, superior to existing deep-based methods and on-par with leading classical (non-deep) sequential and global methods.
format Preprint
id arxiv_https___arxiv_org_abs_2404_14280
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle RESfM: Robust Deep Equivariant Structure from Motion
Khatib, Fadi
Kasten, Yoni
Moran, Dror
Galun, Meirav
Basri, Ronen
Computer Vision and Pattern Recognition
Multiview Structure from Motion is a fundamental and challenging computer vision problem. A recent deep-based approach utilized matrix equivariant architectures for simultaneous recovery of camera pose and 3D scene structure from large image collections. That work, however, made the unrealistic assumption that the point tracks given as input are almost clean of outliers. Here, we propose an architecture suited to dealing with outliers by adding a multiview inlier/outlier classification module that respects the model equivariance and by utilizing a robust bundle adjustment step. Experiments demonstrate that our method can be applied successfully in realistic settings that include large image collections and point tracks extracted with common heuristics that include many outliers, achieving state-of-the-art accuracies in almost all runs, superior to existing deep-based methods and on-par with leading classical (non-deep) sequential and global methods.
title RESfM: Robust Deep Equivariant Structure from Motion
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2404.14280