AI RESEARCH
Disambiguating 2D-3D Correspondences in Gaussian Splatting-based Feature Fields for Visual Localization
arXiv CS.CV
•
ArXi:2605.07351v1 Announce Type: new While Gaussian Splatting-based Feature Fields (GSFFs) have shown promise for visual localization, this paper highlights that photometrically optimized GSFFs are inherently ill-suited for 2D-3D matching. The volumetric extent of each Gaussian induces many-to-one pixel-to-point mappings that destabilize PnP-based pose estimation, while photometric optimization gives rise to superfluous Gaussians devoid of multi-view consistency.