PoseCompass ranks synthetic poses using localization difficulty, coverage novelty, and rendering observability to improve 3DGS-augmented APR fine-tuning, delivering 3x faster adaptation and 53.8% lower median pose errors versus random selection on 7-Scenes.
Dfnet: Enhance absolute pose regression with direct feature matching
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
PoseCompass: Intelligent Synthetic Pose Selection for Visual Localization
PoseCompass ranks synthetic poses using localization difficulty, coverage novelty, and rendering observability to improve 3DGS-augmented APR fine-tuning, delivering 3x faster adaptation and 53.8% lower median pose errors versus random selection on 7-Scenes.