Research map / 3D vision / Human pose estimation
Fusing the Old with the New: Learning Relative Camera Pose with Geometry-Guided Uncertainty
Bingbing Zhuang, Manmohan Chandraker
Check 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 |
|---|---|---|---|
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN Scenes11 | translation error: 3.872 -> 3.001 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN RGB-D | translation error: 16.330 -> 10.520 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN Sun3D | translation error: 12.200 -> 11.830 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN MVS | rotation error: 2.884 -> 2.502 | Hurt |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN MVS | translation error: 5.024 -> 4.506 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN Scenes11 | rotation error: 0.390 -> 0.388 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN RGB-D | rotation error: 1.829 -> 1.480 | Improved |
| Fuse the DNN prediction with 5pt&BA using geometry-guided uncertainty | UA-Fusion · DeMoN Sun3D | rotation error: 1.255 -> 1.340 | Hurt |
| Replace geometry-guided uncertainty with plain aleatoric uncertainty and post-processing fusion (ablation) | UA-Fusion · DeMoN | pose error | Hurt |
| Remove the uncertainty head and fusion; regress pose only (ablation) | UA-Fusion · DeMoN | pose error | Hurt |
| Replace predicted uncertainty with the training-set median geometric uncertainty (ablation) | UA-Fusion · DeMoN | pose error | Hurt |
| Remove self-attention and use plain MLPs for messages (ablation) | UA-Fusion · DeMoN | pose error | 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.
- BA-Net — BA-Net: Dense Bundle Adjustment Networks (ICLR 2019) · 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
- Wide-Baseline Relative Camera Pose Estimation With Directional Learning (CVPR 2021 · similarity 0.90)
- From Correspondences to Pose: Non-minimal Certifiably Optimal Relative Pose without Disambiguation (CVPR 2024 · similarity 0.88)
- FAR: Flexible Accurate and Robust 6DoF Relative Camera Pose Estimation (CVPR 2024 · similarity 0.88)
- Relative Pose from Deep Learned Depth and a Single Affine Correspondence (ECCV 2020 · similarity 0.88)
- DVMNet: Computing Relative Pose for Unseen Objects Beyond Hypotheses (CVPR 2024 · similarity 0.87)
- Relative Pose Estimation through Affine Corrections of Monocular Depth Priors (CVPR 2025 · similarity 0.86)
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.
