Research map / 3D vision / NeRF and radiance fields
Neural Inverse Rendering of an Indoor Scene From a Single Image
Soumyadip Sengupta, Jinwei Gu, Kihwan Kim, Guilin Liu, David W. Jacobs, Jan Kautz
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 |
|---|---|---|---|
| add RAR to the self-supervised reconstruction loss (IIW weak sup.) (ablation) | IRN · CG-PBR albedo | MAD: 0.265 -> 0.130 | Improved |
| add RAR (IIW weak sup.) (ablation) | IRN · IIW albedo | WHDR: 21.6% -> 16.7% | Improved |
| add RAR (IIW weak sup.) (ablation) | IRN · NYUv2 normals | median angular error: 60.2 -> 23.2 | Improved |
| add RAR (IIW weak sup.) (ablation) | IRN · ScanNet normals | median angular error: 91.5 -> 28.4 | Improved |
| add RAR (NYUv2 weak sup.) (ablation) | IRN · CG-PBR albedo | MAD: 0.191 -> 0.134 | Improved |
| add RAR (NYUv2 weak sup.) (ablation) | IRN · IIW albedo | WHDR: 40.6% -> 38.5% | Improved |
| add RAR (NYUv2 weak sup.) (ablation) | IRN · NYUv2 normals | median angular error: 21.1 -> 16.9 | Improved |
| add RAR (NYUv2 weak sup.) (ablation) | IRN · ScanNet normals | median angular error: 87.1 -> 21.1 | Improved |
| add weak supervision over albedo (IIW pairwise judgments) (ablation) | IRN · IIW albedo | WHDR: 32.7% -> 16.7% | Improved |
| add weak supervision over normals (NYUv2) (ablation) | IRN · NYUv2 normals | median angular error: 23.3 -> 16.9 | Improved |
| joint inverse rendering (IRN) instead of single-attribute PBRS for normals | IRN vs PBRS · ScanNet normals | median angular error: 30.39 -> 21.10 | Improved |
| joint inverse rendering (IRN) instead of single-attribute PBRS for normals | IRN vs PBRS · NYUv2 normals | median angular error: 15.33 -> 16.92 | Hurt |
Closest papers by meaning
- Inverse Rendering for Complex Indoor Scenes: Shape, Spatially-Varying Lighting and SVBRDF From a Single Image (CVPR 2020 · similarity 0.92)
- IRISformer: Dense Vision Transformers for Single-Image Inverse Rendering in Indoor Scenes (CVPR 2022 · similarity 0.91)
- Modeling Indirect Illumination for Inverse Rendering (CVPR 2022 · similarity 0.91)
- RobIR: Robust Inverse Rendering for High-Illumination Scenes (NeurIPS 2024 · similarity 0.90)
- Latent Intrinsics Emerge from Training to Relight (NeurIPS 2024 · similarity 0.90)
- Diffusion Renderer: Neural Inverse and Forward Rendering with Video Diffusion Models (CVPR 2025 · similarity 0.89)
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.
