Research map / Graph learning / Graph algorithms and clustering
Efficient Initial Pose-Graph Generation for Global SfM
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use A* traversal with the proposed pose-graph heuristic instead of GC-RANSAC for pose estimation when a graph path yields a valid pose | 1DSfM pose-graph generation · 1DSfM, 402130 view pairs | average pose-estimation runtime: 0.815 -> 0.173 seconds | Helped |
| Use A* traversal rather than breadth-first traversal | Pose-graph pose estimation · 1DSfM, 402130 view pairs | average pose-estimation runtime: 2.916 -> 0.173 seconds | Helped |
| Use Epipolar Hashing to restrict descriptor matching to pose-consistent candidates | 1DSfM feature matching · 1DSfM view pairs | average matching runtime: 0.318 -> 0.057 seconds | Helped |
| Rank correspondences using their inlier history rather than SIFT ratio scores | PROSAC pose estimation · 1DSfM image pairs | average estimation runtime: 0.664 -> 0.615 seconds | Helped |
| Replace exhaustive matching with A* pose estimation and Epipolar Hashing | Global SfM initialization · 1DSfM | total time: 202 -> 29 hours | Helped |
| Use Epipolar Hashing rather than FLANN with A* pose estimation | Global SfM initialization · 1DSfM | multi-view tracks reconstructed: 69312 -> 78335 | Helped |
| Use a minimum spanning tree pose graph instead of exhaustive matching | Global SfM initialization · 1DSfM | views reconstructed: 437 -> 84 | Hurt |
| Add pool-size adaptation to the second-nearest ratio threshold (ablation) | Epipolar Hashing · Madrid Metropolis | views reconstructed: 136 -> 282 | Helped |
| Add pool-size adaptation to the second-nearest ratio threshold (ablation) | Epipolar Hashing · Madrid Metropolis | multi-view tracks reconstructed: 9486 -> 29665 | Helped |
Closest papers by meaning
- Learning to Filter Outlier Edges in Global SfM (CVPR 2025 · similarity 0.89)
- Light3R-SfM: Towards Feed-forward Structure-from-Motion (CVPR 2025 · similarity 0.87)
- Learning Structure-from-Motion with Graph Attention Networks (CVPR 2024 · similarity 0.85)
- VGGSfM: Visual Geometry Grounded Deep Structure From Motion (CVPR 2024 · similarity 0.85)
- From Correspondences to Pose: Non-minimal Certifiably Optimal Relative Pose without Disambiguation (CVPR 2024 · similarity 0.85)
- GRPose: Learning Graph Relations for Human Image Generation with Pose Priors (AAAI 2025 · similarity 0.84)
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.