Get Started
Home
Topics
Search
Library
Research questionHow can neural map matchers learn accurate 3-DoF poses from noisy GPS positions and heading labels?Geo-referenced image datasets often contain noisy position and heading labels, which can mislead models during training. The challenge is to recover useful pose supervision without relying on precisely measured absolute labels.
AI
Computer Vision
Image & Video Processing
Machine Learning
Research Paper
Technology
Latest papersRecent research connected to this question, newest first.AutoCompass: Accurate Visual Localization on Public Maps by Learning from Weak LabelsThe source studies neural map matching with raw GPS labels, tolerance regions around GPS positions, and relative poses from SLAM or SfM when available. Evidence covers driving and egocentric benchmarks.research paper · Sep 2, 2026
Related questions
How can clinical pose estimation remain reliable when preterm infants’ keypoint annotations are noisy?How can HD map construction remain reliable with missing camera views and inaccurate camera poses?How can ground-to-satellite localization bridge severe viewpoint gaps without 3D lifting or depth models?How can inertial measurements recover metric scale in pretrained monocular 3D predictions?