Research map / 3D vision / Depth and stereo
ARC-Loc: Leveraging Azimuthal Ray Convergence as a Geometric Cue for Direct Cross-View Localization
Hyeongsik Kim, Mincheol Kim, Heejoon Moon, Je Hyeong Hong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ray-distance loss (alpha=0.5) to the position loss (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.76 -> 2.52 | Improved |
| raise ray-distance loss weight alpha from 0.5 to 1.0 (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.52 -> 3.98 | Hurt |
| small ray-distance loss weight alpha=0.1 instead of 0.5 (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.76 (alpha=0) vs 2.76 (alpha=0.1) | No clear effect |
| reduce selected matches from N=512 to 256 (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.52 -> 4.57 | Hurt |
| increase selected matches from N=512 to 1024 (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.52 -> 2.67 | Hurt |
| use RADIOv3 backbone instead of DINOv2 (ablation) | ARC-Loc · VIGOR Same-Area | mean localization error (m): 2.90 -> 2.52 m; latency 94.53 -> 59.67 ms | Improved |
| direct 2D matching with azimuthal-ray solver instead of depth-based lifting (ARC-Loc vs Loc2) | Non-BEV cross-view localization · VIGOR Same-Area | mean localization error (m): 3.06 -> 2.52 | Improved |
| azimuthal-ray solver vs depth-based Loc2 under restricted FoV | ARC-Loc · KITTI Cross-Area | mean localization error (m): Loc2 5.60 vs ARC-Loc 7.45 | Hurt |
| remove BEV construction and depth module (ARC-Loc vs FG2 and Loc2) | ARC-Loc · VIGOR inference | matching latency (ms): 187.61 (FG2) / 137.98 (Loc2) -> 58.69 | Improved |
| ARC-Loc vs BEV-based FG2 on VIGOR known orientation | ARC-Loc · VIGOR Same-Area | mean localization error (m): FG2 1.95 vs ARC-Loc 2.52 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Loc2 — Loc$^{2}$: Interpretable Cross-View Localization via Depth-Lifted Local Feature Matching (ICLR 2026) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Beyond 2D Matching: A Unified Single-Stage Framework for Geometry-Aware Cross-View Object Geo-Localization (ECCV 2026 · similarity 0.90)
- Seeing Across Skies and Streets: Feedforward 3D Reconstruction from Satellite, Drone, and Ground Images (NeurIPS 2026 · similarity 0.89)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.87)
- BevSplat: Resolving Height Ambiguity via Feature-Based Gaussian Primitives for Weakly-Supervised Cross-View Localization (NeurIPS 2025 · similarity 0.87)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.87)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
