Research map / 3D vision / Depth and stereo
Variational Uncalibrated Photometric Stereo Under General Lighting
Bjoern Haefner, Zhenzhang Ye, Maolin Gao, Tao Wu, Yvain Queau, Daniel Cremers
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 |
|---|---|---|---|
| use minimal-surface balloon initialization instead of hemisphere initialization (ablation) | Variational UPS (ours) · Synthetic UPS (36 datasets) | Median MAE: 59.41 -> 9.17 | Improved |
| use minimal-surface balloon initialization instead of hemisphere initialization (ablation) | Variational UPS (ours) · Synthetic UPS (36 datasets) | Mean MAE: 55.20 -> 10.72 | Improved |
| use minimal-surface balloon initialization instead of initializing from a prior method (ablation) | Variational UPS (ours) · Synthetic UPS (36 datasets) | Median MAE: 19.86 -> 9.17 | Improved |
| use minimal-surface balloon initialization instead of initializing from a prior method (ablation) | Variational UPS (ours) · Synthetic UPS (36 datasets) | Mean MAE: 27.43 -> 10.72 | Improved |
| balloon initialization instead of prior-method initialization on Lucy with Hippie albedo (ablation) | Variational UPS (ours) · Lucy, Hippie albedo | Mean angular error: 8.10 -> 7.87 | No clear effect |
| balloon initialization instead of prior-method initialization on Armadillo with Hippie albedo (ablation) | Variational UPS (ours) · Armadillo, Hippie albedo | Mean angular error: 12.56 -> 14.52 | Hurt |
| balloon initialization instead of prior-method initialization on Thai Statue with Rectcircle albedo (ablation) | Variational UPS (ours) · Thai Statue, Rectcircle albedo | Mean angular error: 79.88 -> 8.84 | Improved |
| increase the number of input images M (15 to 25 is a good trade-off) (ablation) | Variational UPS (ours) · Synthetic datasets | Mean angular error | Improved |
| increase the number of input images M (ablation) | Variational UPS (ours) · Synthetic datasets | Runtime | Hurt |
| use too small or too large Huber-TV weight mu instead of mu in [1e-6, 1e-5] (ablation) | Variational UPS (ours) · Synthetic datasets | Depth accuracy | Hurt |
| full method versus best prior method (column 2) | Variational UPS (ours) · Synthetic UPS (36 datasets) | Median MAE: 21.14 -> 9.17 | Improved |
Closest papers by meaning
- A Non-Convex Variational Approach to Photometric Stereo Under Inaccurate Lighting (CVPR 2017 · similarity 0.92)
- A Benchmark Dataset and Evaluation for Non-Lambertian and Uncalibrated Photometric Stereo (CVPR 2016 · similarity 0.91)
- Semi-Calibrated Near Field Photometric Stereo (CVPR 2017 · similarity 0.90)
- Deep near-light photometric stereo for spatially varying reflectances (ECCV 2020 · similarity 0.90)
- Uncalibrated Photometric Stereo by Stepwise Optimization Using Principal Components of Isotropic BRDFs (CVPR 2016 · similarity 0.90)
- Sparse Views Near Light: A Practical Paradigm for Uncalibrated Point-light Photometric Stereo (CVPR 2024 · 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.
