Research map / 3D vision / Depth and stereo
Very Large-Scale Global SfM by Distributed Motion Averaging
Read the paperCheck 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 distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-A | peak memory (GB): 134.01 -> 7.06 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-A | time (hours): 86.70 -> 9.44 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-B | peak memory (GB): 13.76 -> 0.72 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-B | time (hours): 17.85 -> 2.45 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-C | peak memory (GB): 9.12 -> 0.38 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-C | time (hours): 10.95 -> 1.95 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-D | peak memory (GB): 3.62 -> 0.28 | Helped |
| Use distributed divide-and-conquer motion averaging instead of traditional motion averaging | Global SfM · City-D | time (hours): 3.65 -> 0.95 | Helped |
| Use 32 partitions rather than a single partition (ablation) | Distributed motion averaging · City-scale data-sets | time: 15% of the time | Helped |
| Remove intra-camera variables and merge partitions only by similarity transformations (ablation) | Distributed motion averaging · Illustrated reconstruction | camera-pose quality | Hurt |
| Add intra-camera variables to the partition optimization (ablation) | Distributed motion averaging · Illustrated reconstruction | camera-pose quality | Helped |
| Increase relative-motion Gaussian noise (ablation) | Distributed versus traditional motion averaging · Temple of Heaven | close-loop reconstruction: 7σr and 16σt | Mixed |
Closest papers by meaning
- Light3R-SfM: Towards Feed-forward Structure-from-Motion (CVPR 2025 · similarity 0.86)
- Dense-SfM: Structure from Motion with Dense Consistent Matching (CVPR 2025 · similarity 0.85)
- ColabSfM: Collaborative Structure-from-Motion by Point Cloud Registration (CVPR 2025 · similarity 0.85)
- Learning to Filter Outlier Edges in Global SfM (CVPR 2025 · similarity 0.84)
- MoDec-GS: Global-to-Local Motion Decomposition and Temporal Interval Adjustment for Compact Dynamic 3D Gaussian Splatting (CVPR 2025 · similarity 0.84)
- 4D Gaussian Splatting in the Wild with Uncertainty-Aware Regularization (NeurIPS 2024 · 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.