AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

Learning Correspondence Uncertainty via Differentiable Nonlinear Least Squares

Dominik Muhle, Lukas Koestler, Krishna Murthy Jatavallabhula, Daniel Cremers

CVPR 2023 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Priors and functional variational uncertainty · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
learn feature-position covariances using differentiable pose optimizationPNEC relative pose estimation · KITTI test, SuperPointRPE_n: 5.353 -> 3.333Improved
learn feature-position covariances using differentiable pose optimizationPNEC relative pose estimation · KITTI test, KLT tracksRPE_n: 3.060 -> 2.359Improved
use self-supervised cycle-consistency training instead of supervised pose traininglearned covariance model · KITTI test, KLT tracksRPE_n: 2.359 -> 1.910Improved
use a KITTI-trained self-supervised model on EuRoClearned covariance model · EuRoC mean, SuperPointRPE_n: 58.90 -> 30.57Improved
use varied pose geometries during covariance learning (ablation)synthetic covariance learning · synthetic experiments2D covariance estimationImproved
use a single shared geometric setup rather than diverse pose geometries (ablation)synthetic covariance learning · synthetic experiments2D covariance estimationHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy