English

High Fidelity 3D Reconstructions with Limited Physical Views

Computer Vision and Pattern Recognition 2022-10-06 v1 Artificial Intelligence Machine Learning Robotics

Abstract

Multi-view triangulation is the gold standard for 3D reconstruction from 2D correspondences given known calibration and sufficient views. However in practice, expensive multi-view setups -- involving tens sometimes hundreds of cameras -- are required in order to obtain the high fidelity 3D reconstructions necessary for many modern applications. In this paper we present a novel approach that leverages recent advances in 2D-3D lifting using neural shape priors while also enforcing multi-view equivariance. We show how our method can achieve comparable fidelity to expensive calibrated multi-view rigs using a limited (2-3) number of uncalibrated camera views.

Keywords

Cite

@article{arxiv.2110.11599,
  title  = {High Fidelity 3D Reconstructions with Limited Physical Views},
  author = {Mosam Dabhi and Chaoyang Wang and Kunal Saluja and Laszlo Jeni and Ian Fasel and Simon Lucey},
  journal= {arXiv preprint arXiv:2110.11599},
  year   = {2022}
}

Comments

Accepted to 3DV 2021 (project page & code: https://sites.google.com/view/high-fidelity-3d-neural-prior)