Research map / 3D vision / Depth and stereo
Learning 3D Reconstruction with Priors in Test Time
Lei Zhou, Haoyu Wu, Akshat Dave, Dimitris Samaras
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 compatibility objective with no priors | TCO-VGGT · ETH3D | Acc. mean: 0.280 -> 0.270 | Improved |
| add intrinsics prior to compatibility objective (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.270 -> 0.273 | No clear effect |
| add camera pose prior to compatibility objective | TCO-VGGT · ETH3D | Acc. mean: 0.270 -> 0.185 | Improved |
| add intrinsics prior on top of pose and compatibility (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.185 -> 0.114 | Improved |
| add compatibility objective to pose and intrinsics priors (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.134 -> 0.114 | Improved |
| use raw VGGT depth with prior camera parameters (ablation) | VGGT · ETH3D | Acc. mean: 0.280 -> 0.283 | No clear effect |
| add pose and intrinsics priors without compatibility objective (ablation) | TCO-VGGT · DTU | Acc. mean / Comp. mean: Acc. 1.338 -> 1.327, Comp. 1.896 -> 1.968 | Mixed |
| add compatibility objective to pose and intrinsics priors (ablation) | TCO-VGGT · DTU | Acc. mean: 1.327 -> 0.976 | Improved |
| add compatibility objective with no priors | TCO-VGGT · DTU | Acc. mean: 1.338 -> 1.170 | Improved |
| also fine-tune the camera head (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.114 -> 0.158 | Hurt |
| also fine-tune the depth head (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.114 -> 0.162 | Hurt |
| use confidence as 2DGS opacity (ablation) | TCO-VGGT · ETH3D | Acc. mean: 0.120 -> 0.114 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 7 of 7 shared results
- FLARE — FLARE: Feed-forward Geometry, Appearance and Camera Estimation from Uncalibrated Sparse Views (CVPR 2025) · better on 1 of 1 shared result
- Fast3R — Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass (CVPR 2025) · better on 1 of 1 shared result
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
- Test3R: Learning to Reconstruct 3D at Test Time (NeurIPS 2025 · similarity 0.90)
- Déjà View: Looping Transformers for Multi-View 3D Reconstruction (NeurIPS 2026 · similarity 0.88)
- Understanding Multi-View Transformers (NeurIPS 2026 · similarity 0.88)
- The More You See in 2D the More You Perceive in 3D (CVPR 2024 · similarity 0.88)
- VGGT: Visual Geometry Grounded Transformer (CVPR 2025 · similarity 0.88)
- Pow3R: Empowering Unconstrained 3D Reconstruction with Camera and Scene Priors (CVPR 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.
