Research map / 3D vision / Depth and stereo
Deep Photometric Stereo for Non-Lambertian Surfaces
Guanying Chen, Kai Han, Boxin Shi, Yasuyuki Matsushita, Kenneth K. Wong
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 conv fusion with max-pooling fusion (ablation) | PS-FCN · SynTest_MERL Sphere | mean angular error: 4.54 -> 3.65 | Improved |
| replace average-pooling with max-pooling fusion (ablation) | PS-FCN · SynTest_MERL Sphere | mean angular error: 3.71 -> 3.40 | Improved |
| add sculpture dataset to blobby training data (ablation) | PS-FCN · SynTest_MERL Dragon | mean angular error: 7.23 -> 4.83 | Improved |
| test with 100 images instead of 32 (ablation) | PS-FCN · SynTest_MERL Sphere | mean angular error: 3.65 -> 3.40 | Improved |
| use clustered lighting (22 degree range) instead of diverse lighting (144 degree) (ablation) | PS-FCN · Bunny | mean angular error: 4.21 -> 18.72 | Hurt |
| add data normalization (+N) (ablation) | PS-FCN · DiLiGenT (calibrated) | mean angular error: 8.4 -> 7.4 | Improved |
| add object mask input to LCNet (ablation) | LCNet · SynTest_MERL Bunny | light direction MAE (deg): 8.85 -> 5.38 | Improved |
| add local-global feature fusion to LCNet (ablation) | LCNet · SynTest_MERL Sphere | light direction MAE (deg): 6.87 -> 3.47 | Improved |
| classify light direction in bins instead of regressing (ablation) | LCNet · DiLiGenT light direction | light direction MAE (deg): 6.68 -> 4.92 | Improved |
| classify light direction in bins instead of regressing (ablation) | LCNet + PS-FCN dagger · DiLiGenT (uncalibrated) | mean angular error: 11.0 -> 9.5 | Improved |
| train PS-FCN on lightings estimated by LCNet (PS-FCN dagger) (ablation) | LCNet + PS-FCN · SynTest_MERL Sphere | mean angular error: 3.19 -> 2.71 | Improved |
| integrate LCNet with PS-FCN+N (ablation) | LCNet + PS-FCN · SynTest_MERL Sphere | mean angular error: 3.19 -> 4.53 | Hurt |
Closest papers by meaning
- PS-FCN: A Flexible Learning Framework for Photometric Stereo (ECCV 2018 · similarity 0.97)
- Deep Learning Methods for Calibrated Photometric Stereo and Beyond (TPAMI 2024 · similarity 0.94)
- CNN-PS: CNN-based Photometric Stereo for General Non-Convex Surfaces (ECCV 2018 · similarity 0.93)
- Deep Photometric Stereo Networks for Determining Surface Normal and Reflectances (TPAMI 2020 · similarity 0.93)
- Self-Calibrating Deep Photometric Stereo Networks (CVPR 2019 · similarity 0.92)
- A CNN Based Approach for the Point-Light Photometric Stereo Problem (IJCV 2022 · 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.
