Research map / 3D vision / 3D generation
Object-Driven Multi-Layer Scene Decomposition From a Single Image
Helisa Dhamo, Nassir Navab, Federico Tombari
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 class scores (ablation) | Ours · SunCG | 1st layer depth MPE: 0.551 -> 0.508 | Improved |
| add class scores (ablation) | Ours · SunCG | 2nd layer depth MPE: 0.687 -> 0.700 | Hurt |
| add class scores (ablation) | Ours · SunCG | 2nd layer RGB MPE: 43.97 -> 44.50 | Hurt |
| add perceptual loss Lp (ablation) | Ours · SunCG | 2nd layer depth MPE: 0.700 -> 0.657 | Improved |
| add perceptual loss Lp (ablation) | Ours · SunCG | 2nd layer RGB MPE: 44.50 -> 43.92 | Improved |
| add re-composition loss (ablation) | Ours · SunCG | 1st layer depth MPE: 0.496 -> 0.473 | Improved |
| add re-composition loss (ablation) | Ours · SunCG | 2nd layer depth MPE: 0.657 -> 0.641 | Improved |
| add re-composition loss (ablation) | Ours · Stanford 2D-3D | 1st layer depth MPE: 0.509 -> 0.469 | Improved |
| add re-composition loss (ablation) | Ours · Stanford 2D-3D | 2nd layer depth MPE: 0.692 -> 0.688 | Improved |
| add input depth prediction to layout branch (ablation) | Network B · SunCG layout | depth MPE: 0.662 -> 0.505 | Improved |
| add input depth prediction to layout branch (ablation) | Network B · SunCG layout | color MPE: 21.42 -> 22.64 | Hurt |
| add adversarial loss (ablation) | Network B · SunCG layout | color MPE: 22.64 -> 20.93 | Improved |
Closest papers by meaning
- Layered Scene Decomposition via the Occlusion-CRF (CVPR 2016 · similarity 0.88)
- 3D Photography Using Context-Aware Layered Depth Inpainting (CVPR 2020 · similarity 0.88)
- $\alpha$Depth: Learning Single-Pass Soft Boundary Decomposition for Stereo Conversion (NeurIPS 2026 · similarity 0.86)
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 2024 · similarity 0.85)
- O^2-Recon: Completing 3D Reconstruction of Occluded Objects in the Scene with a Pre-trained 2D Diffusion Model (AAAI 2024 · similarity 0.85)
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 2025 · similarity 0.85)
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.
