Research map / 3D vision / Depth and stereo
Surface Normal Estimation of Tilted Images via Spatial Rectifier
Tien Do, Khiem Vuong, Stergios I. Roumeliotis, Hyun Soo Park
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 spatial rectifier to DFPN+TAL (ablation) | DFPN+TAL · Tilt-RGBD tilted images | mean angular error (deg): 14.53 -> 11.22 | Improved |
| add spatial rectifier to DORN+TAL (ablation) | DORN+TAL · Tilt-RGBD tilted images | mean angular error (deg): 14.33 -> 11.77 | Improved |
| add spatial rectifier to DFPN+TAL (ablation) | DFPN+TAL · Tilt-RGBD gravity-aligned images | mean angular error (deg): 9.74 -> 8.59 | Improved |
| add spatial rectifier rather than tilt data augmentation (ablation) | DFPN+TAL · Tilt-RGBD tilted images | mean angular error (deg): 12.89 -> 11.22 | Improved |
| add tilt data augmentation (ablation) | DFPN+TAL · Tilt-RGBD tilted images | mean angular error (deg): 14.53 -> 12.89 | Improved |
| add tilt data augmentation (ablation) | DFPN+TAL · Tilt-RGBD gravity-aligned images | mean angular error (deg): 9.74 -> 10.19 | Hurt |
| use learned spatial rectifier instead of IMU gravity (ablation) | DFPN+TAL · Tilt-RGBD tilted images | mean angular error (deg): 11.46 -> 11.22 | No clear effect |
| use DFPN instead of DORN with spatial rectifier (ablation) | spatial rectifier · Tilt-RGBD tilted images | mean angular error (deg): 11.77 -> 11.22 | Improved |
| use ResNeXt-101 instead of ResNet-101 backbone (ablation) | DFPN+TAL · ScanNet | mean angular error (deg): 15.4 -> 15.0 | Improved |
| use ResNeXt-101 instead of ResNet-101 backbone (ablation) | DFPN+TAL · NYUv2 | mean angular error (deg): 16.9 -> 16.2 | Improved |
| use DFPN instead of Panoptic FPN (ablation) | TAL training · ScanNet | mean angular error (deg): 15.8 -> 15.4 | Improved |
| use truncated angular loss instead of L2 (ablation) | DFPN ResNeXt-101 · NYUv2 | mean angular error (deg): 17.1 -> 16.2 | Improved |
Closest papers by meaning
- Rethinking Inductive Biases for Surface Normal Estimation (CVPR 2024 · similarity 0.87)
- Adaptive Surface Normal Constraint for Geometric Estimation From Monocular Images (TPAMI 2024 · similarity 0.86)
- Estimating and Exploiting the Aleatoric Uncertainty in Surface Normal Estimation (ICCV 2021 · similarity 0.86)
- Poppy: Polarization-Based Plug-and-Play Guidance for Enhancing Surface Normal Estimation (ECCV 2026 · similarity 0.86)
- Fast Globally Optimal Surface Normal Estimation from an Affine Correspondence (ICCV 2023 · similarity 0.86)
- Normal-NeRF: Ambiguity-Robust Normal Estimation for Highly Reflective Scenes (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.
