Research map / 3D vision / Depth and stereo
Adaptive Surface Normal Constraint for Depth Estimation
Xiaoxiao Long, Cheng Lin, Lingjie Liu, Wei Li, Christian Theobalt, Ruigang Yang, Wenping Wang
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 |
|---|---|---|---|
| replace L1-only training with L1 plus ASN constraint (ablation) | depth network (HRNet-18) · NYUD-V2 | recovered normal mean angle error: 31.3 -> 22.2 | Improved |
| use ASN instead of Sobel-like normal constraint (ablation) | depth network (HRNet-18) · NYUD-V2 | rel: 0.118 -> 0.111 | Improved |
| use ASN instead of least-square normal constraint (ablation) | depth network (HRNet-18) · NYUD-V2 | rel: 0.119 -> 0.111 | Improved |
| add Sobel-like normal constraint to L1 loss (ablation) | depth network (HRNet-18) · NYUD-V2 | rel: 0.113 -> 0.118 | Hurt |
| add least-square normal constraint to L1 loss (ablation) | depth network (HRNet-18) · NYUD-V2 | rel: 0.113 -> 0.119 | Hurt |
| add virtual normal constraint to L1 loss (ablation) | depth network (HRNet-18) · NYUD-V2 | recovered normal mean angle error: 31.3 -> 31.7 | No clear effect |
| use ASN instead of virtual normal (ablation) | depth network (HRNet-18) · NYUD-V2 | recovered normal mean angle error: 31.7 -> 22.2 | Improved |
| add geometric context adaption to area adaption (ablation) | ASN · NYUD-V2 | recovered normal mean angle error: 22.6 -> 22.2 | Improved |
| add area adaption to geometric context adaption (ablation) | ASN · NYUD-V2 | recovered normal mean angle error: 22.3 -> 22.2 | Improved |
| increase local patch size from 3 to 7 (ablation) | ASN · NYUD-V2 | recovered normal mean angle error: 22.5 -> 22.2 | No clear effect |
| increase number of sampled triplets (ablation) | ASN · NYUD-V2 | recovered normal accuracy | Improved |
| use area weighting instead of simple average of candidates (ablation) | ASN · synthetic noisy semi-sphere | mean angle error | Improved |
Closest papers by meaning
- Adaptive Surface Normal Constraint for Geometric Estimation From Monocular Images (TPAMI 2024 · similarity 0.96)
- Enforcing Geometric Constraints of Virtual Normal for Depth Prediction (ICCV 2019 · similarity 0.90)
- Virtual Normal: Enforcing Geometric Constraints for Accurate and Robust Depth Prediction (TPAMI 2021 · similarity 0.90)
- Normal Assisted Stereo Depth Estimation (CVPR 2020 · similarity 0.89)
- Unsupervised Learning of Geometry From Videos With Edge-Aware Depth-Normal Consistency (AAAI 2018 · similarity 0.89)
- CORE: Co-planarity Regularized Monocular Geometry Estimation with Weak Supervision (ICCV 2023 · similarity 0.88)
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.
