$L^3$:Scene-agnostic Visual Localization in the Wild

2026-03-09

AI Summary

This work addresses the reliance of visual localization on offline 3D mapping by proposing the first general-purpose localization framework that requires no pre-built scene representation. The method leverages a feedforward 3D reconstruction network to directly recover metric-scale 3D structure from a single RGB image in an online manner, combined with a two-stage strategy for scale recovery and pose refinement to achieve real-time, high-precision localization in arbitrary outdoor scenes. Evaluated on multiple benchmark datasets, the approach attains accuracy comparable to state-of-the-art methods while demonstrating significantly improved robustness in challenging scenarios with sparse reference imagery.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Representation Learning for Vision
• Machine Learning: Representation Learning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Location- and context-aware Web and WoT applications and services

Abstract

Standard visual localization methods typically require offline pre-processing of scenes to obtain 3D structural information for better performance. This inevitably introduces additional computational and time costs, as well as the overhead of storing scene representations. Can we visually localize in a wild scene without any off-line preprocessing step? In this paper, we leverage the online inference capabilities of feed-forward 3D reconstruction networks to propose a novel map-free visual localization framework $L^3$. Specifically, by performing direct online 3D reconstruction on RGB images, followed by two-stage metric scale recovery and pose refinement based on 2D-3D correspondences, $L^3$ achieves high accuracy without the need to pre-build or store any offline scene representations. Extensive experiments demonstrate $L^3$ not only that the performance is comparable to state-of-the-art solutions on various benchmarks, but also that it exhibits significantly superior robustness in sparse scenes (fewer reference images per scene).

Problem

• visual localization
• scene-agnostic
• map-free
• 3D reconstruction
• wild scenes

Innovation

• map-free localization
• online 3D reconstruction
• scene-agnostic
• visual localization
• 2D-3D correspondence