Research map / 3D vision / Depth and stereo
Learning to Minify Photometric Stereo
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| learn and select informative illumination directions with the sparse connection table | photometric stereo · DiLiGenT, 10 inputs | MAE: 14.34 -> 10.02 | Helped |
| learn and select informative illumination directions with the sparse connection table | photometric stereo · DiLiGenT, 8 inputs | MAE: 11.42 -> 10.39 | Helped |
| reduce input directions from 96 to 16 using learned selection | Ours-g(·) · DiLiGenT | MAE: 8.43 -> 9.66 | Hurt |
| use g(·) regularization instead of the L1 regularizer (ablation) | proposed connection-table method · DiLiGenT, 10 inputs | MAE: 11.33 -> 10.02 | Helped |
| pre-train and regularize the connection table instead of randomly initializing it without regularization (ablation) | proposed connection-table method · DiLiGenT, 10 inputs | MAE: 11.85 -> 10.02 | Helped |
| add occlusion-layer augmentation to 5% of training maps (ablation) | photometric stereo network · Synthetic training data, 10 inputs | MAE: 11.42 -> 10.42 | Helped |
| add occlusion-layer augmentation to 5% of training maps (ablation) | photometric stereo network · Real-world training data, 10 inputs | MAE: 8.28 -> 7.99 | Helped |
| use structured occlusion-layer masks instead of random zeroing at 5% (ablation) | photometric stereo network · Synthetic training data, 10 inputs | MAE: 10.87 -> 10.42 | Helped |
| use structured occlusion-layer masks instead of random zeroing at 5% (ablation) | photometric stereo network · Real-world training data, 10 inputs | MAE: 8.13 -> 7.99 | Helped |
Closest papers by meaning
- Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018 · similarity 0.90)
- Universal Photometric Stereo Network Using Global Lighting Contexts (CVPR 2022 · similarity 0.90)
- Uncalibrated Neural Inverse Rendering for Photometric Stereo of General Surfaces (CVPR 2021 · similarity 0.90)
- GPS-Net: Graph-based Photometric Stereo Network (NeurIPS 2020 · similarity 0.90)
- Sparse Views Near Light: A Practical Paradigm for Uncalibrated Point-light Photometric Stereo (CVPR 2024 · similarity 0.86)
- Differentiable Display Photometric Stereo (CVPR 2024 · 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.