Research map / 3D vision / Depth and stereo
CNN-PS: CNN-based Photometric Stereo for General Non-Convex Surfaces
Satoshi Ikehata
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 |
|---|---|---|---|
| average predictions over rotated observation maps (K=10 instead of K=1) at test time (ablation) | CNN-PS · CyclesPSTest | mean angular error | Improved |
| use observation map width w=32 rather than 8, 16 or 64 (ablation) | CNN-PS · normal estimation with fewer than one thousand images | angular error | Improved |
| smoothly interpolate missing entries of the observation map (ablation) | CNN-PS · normal estimation | performance | Hurt |
| add batch normalization layers to the network (ablation) | CNN-PS · normal estimation | performance | Hurt |
| project lights in spherical coordinates (theta, phi) instead of the x-y plane (ablation) | CNN-PS · normal estimation | performance | Hurt |
| swap DenseNet for AlexNet, VGG or shallow conv networks (ablation) | CNN-PS · normal estimation | performance | No clear effect |
| train on CyclesPS with global illumination rendering and compare with robust and model-based methods | CNN-PS · CyclesPSTest non-convex objects (TURTLE, PAPERBOWL) | mean angular error | Improved |
| compare CNN-PS against IA14 on rendered MERL BRDF spheres | CNN-PS · MERLSphere | mean angular error | No clear effect |
Closest papers by meaning
- A CNN Based Approach for the Point-Light Photometric Stereo Problem (IJCV 2022 · similarity 0.93)
- PS-FCN: A Flexible Learning Framework for Photometric Stereo (ECCV 2018 · similarity 0.93)
- Deep Photometric Stereo for Non-Lambertian Surfaces (TPAMI 2020 · similarity 0.93)
- Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018 · similarity 0.92)
- Deep Learning Methods for Calibrated Photometric Stereo and Beyond (TPAMI 2024 · similarity 0.92)
- Uncalibrated Neural Inverse Rendering for Photometric Stereo of General Surfaces (CVPR 2021 · similarity 0.92)
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.
