GSVisLoc: Generalizable Visual Localization for Gaussian Splatting Scene Representations
By: Fadi Khatib , Dror Moran , Guy Trostianetsky and more
Potential Business Impact:
Finds where a camera is in a 3D scene.
We introduce GSVisLoc, a visual localization method designed for 3D Gaussian Splatting (3DGS) scene representations. Given a 3DGS model of a scene and a query image, our goal is to estimate the camera's position and orientation. We accomplish this by robustly matching scene features to image features. Scene features are produced by downsampling and encoding the 3D Gaussians while image features are obtained by encoding image patches. Our algorithm proceeds in three steps, starting with coarse matching, then fine matching, and finally by applying pose refinement for an accurate final estimate. Importantly, our method leverages the explicit 3DGS scene representation for visual localization without requiring modifications, retraining, or additional reference images. We evaluate GSVisLoc on both indoor and outdoor scenes, demonstrating competitive localization performance on standard benchmarks while outperforming existing 3DGS-based baselines. Moreover, our approach generalizes effectively to novel scenes without additional training.
Similar Papers
Gaussian Splatting Feature Fields for Privacy-Preserving Visual Localization
CV and Pattern Recognition
Helps cameras find their exact spot in a place.
3D Gaussian Splatting aided Localization for Large and Complex Indoor-Environments
CV and Pattern Recognition
Makes robots see better in tricky places.
GSFeatLoc: Visual Localization Using Feature Correspondence on 3D Gaussian Splatting
CV and Pattern Recognition
Finds exactly where a photo was taken in 3D.