Research map / 3D vision / Depth and stereo
Long-Tail Internet Photo Reconstruction
Yuan Li, Yuanbo Xiangli, Hadar Averbuch-Elor, Noah Snavely, Ruojin Cai
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 |
|---|---|---|---|
| finetune on MegaDepth-X with sparsity-aware sampling | pi3 · MegaDepth-X (hard) | AUC@5: 36.93 -> 47.93 | Improved |
| finetune on MegaDepth-X with sparsity-aware sampling | VGGT · MegaDepth-X (hard) | AUC@5: 29.10 -> 41.49 | Improved |
| finetune on MegaDepth-X with sparsity-aware sampling | pi3 · MegaDepth-X (easy) | AUC@5: 45.84 -> 55.58 | Improved |
| finetune on MegaDepth-X with sparsity-aware sampling | pi3 · MegaDepth-X (hard) | MRE: 12.21 -> 5.72 | Improved |
| add filtering and doppelganger disambiguation of training data (vs Dirty) (ablation) | pi3-FT · MegaDepth-X (hard) | AUC@5: 43.74 -> 47.93 | Improved |
| add filtering and doppelganger disambiguation of training data (vs Dirty) (ablation) | pi3-FT · MegaDepth-X (hard) | Acc (mean): 0.130 -> 0.068 | Improved |
| use Mixed sampling instead of Random (ablation) | pi3-FT · MegaDepth-X (hard) | MRE: 6.53 -> 5.72 | Improved |
| use Mixed sampling instead of Dense (ablation) | pi3-FT · MegaDepth-X (hard) | MRE: 6.04 -> 5.72 | Improved |
| use Mixed sampling instead of Sparse (ablation) | pi3-FT · MegaDepth-X (easy) | AUC@5: 55.12 -> 55.58 | Improved |
| use Mixed sampling instead of Sparse (ablation) | pi3-FT · MegaDepth-X (hard) | AUC@5: 47.13 -> 47.93 | No clear effect |
| finetune on MegaDepth-X | VGGT · RealEstate10K | AUC@5: 38.09 -> 48.23 | Improved |
| finetune on MegaDepth-X | pi3 · RealEstate10K | AUC@5: 62.82 -> 60.01 | 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.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 7 of 7 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
- MegaSynth: Scaling Up 3D Scene Reconstruction with Synthesized Data (CVPR 2025 · similarity 0.85)
- Sparse-view Pose Estimation and Reconstruction via Analysis by Generative Synthesis (NeurIPS 2024 · similarity 0.85)
- Extreme Rotation Estimation in the Wild (CVPR 2025 · similarity 0.85)
- Sparse3D: Distilling Multiview-Consistent Diffusion for Object Reconstruction from Sparse Views (AAAI 2024 · similarity 0.85)
- Stereo4D: Learning How Things Move in 3D from Internet Stereo Videos (CVPR 2025 · similarity 0.84)
- Generating 3D-Consistent Videos from Unposed Internet Photos (CVPR 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.
