Research map / 3D vision / Depth and stereo
Self-Calibrating Deep Photometric Stereo Networks
Guanying Chen, Kai Han, Boxin Shi, Yasuyuki Matsushita, Kwan-Yee K. Wong
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 |
|---|---|---|---|
| use a two-stage lighting-calibration and normal-estimation pipeline | uncalibrated photometric stereo · DiLiGenT benchmark | normal MAE (degrees, Avg.): 13.62 -> 9.51 | Improved |
| replace direct lighting regression with classification (ablation) | LCNet · MERL Test Sphere | light direction MAE (degrees): 4.10 -> 3.47 | Improved |
| add the object mask as input (ablation) | LCNet · MERL Test Sphere | light direction MAE (degrees): 5.46 -> 3.47 | Improved |
| add local-global feature fusion across observations (ablation) | LCNet · MERL Test Sphere | light direction MAE (degrees): 6.87 -> 3.47 | Improved |
| train NENet using classification-estimated rather than regression-estimated lighting (ablation) | normal estimation · MERL Test Bunny | normal MAE (degrees): 4.99 -> 4.09 | Improved |
| use NENet rather than PS-FCN with LCNet-estimated lighting (ablation) | normal estimation · MERL Test Bunny | normal MAE (degrees): 4.67 -> 4.09 | Improved |
| fine-tune the two stages end to end (ablation) | SDPS-Net · not specified | performance | No clear effect |
| increase the depth of the single-stage UPS-FCN model (ablation) | UPS-FCN · MERL Test Sphere | normal MAE (degrees): 7.44 -> 4.30 | Mixed |
| increase the width of the single-stage UPS-FCN model (ablation) | UPS-FCN · MERL Test Sphere | normal MAE (degrees): 7.44 -> 5.61 | Hurt |
Closest papers by meaning
- Deep Learning Methods for Calibrated Photometric Stereo and Beyond (TPAMI 2024 · similarity 0.93)
- Deep Photometric Stereo for Non-Lambertian Surfaces (TPAMI 2020 · similarity 0.92)
- PS-FCN: A Flexible Learning Framework for Photometric Stereo (ECCV 2018 · similarity 0.92)
- What is Learned in Deep Uncalibrated Photometric Stereo? (ECCV 2020 · similarity 0.91)
- CNN-PS: CNN-based Photometric Stereo for General Non-Convex Surfaces (ECCV 2018 · similarity 0.91)
- Uncalibrated Neural Inverse Rendering for Photometric Stereo of General Surfaces (CVPR 2021 · similarity 0.90)
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.
