Research map / 3D vision / 3D generation
SPSG: Self-Supervised Photometric Scene Generation From RGB-D Scans
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 |
|---|---|---|---|
| use 2D view-guided rendering losses instead of 3D target supervision (ablation) | SPSG · Matterport3D | IoU: 0.33 -> 0.39 | Helped |
| use 2D view-guided rendering losses instead of 3D target supervision (ablation) | SPSG · Matterport3D | FID: 80.51 -> 56.03 | Helped |
| add adversarial and perceptual losses to the ℓ1 reconstruction loss (ablation) | SPSG · Matterport3D | FID: 67.92 -> 56.03 | Helped |
| add adversarial loss (ablation) | SPSG · Matterport3D | FID: 62.15 -> 56.03 | Helped |
| add perceptual loss (ablation) | SPSG · Matterport3D | FID: 61.46 -> 56.03 | Helped |
| use CIELAB instead of RGB color space (ablation) | SPSG · Matterport3D | FID: 58.8 -> 56.03 | Helped |
| use CIELAB instead of RGB color space (ablation) | SPSG · Matterport3D | Feature-ℓ1: 0.222 -> 0.219 | Helped |
| use 2D view-guided losses instead of a 3D ℓ1 loss only (ablation) | SPSG · Matterport3D | IoU: 0.31 -> 0.39 | Helped |
| train on real Matterport3D data instead of synthetic data | SG-NN · Matterport3D | IoU: 0.27 -> 0.28 | Helped |
Closest papers by meaning
- SG-NN: Sparse Generative Neural Networks for Self-Supervised Scene Completion of RGB-D Scans (CVPR 2020 · similarity 0.89)
- 3D Gaussian Splatting within Self-Learned Neural View-Dependent Color Fields (NeurIPS 2026 · similarity 0.87)
- Adversarial Texture Optimization From RGB-D Scans (CVPR 2020 · similarity 0.87)
- TransformerFusion: Monocular RGB Scene Reconstruction using Transformers (NeurIPS 2021 · similarity 0.86)
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 2024 · similarity 0.86)
- ERUPT: Efficient Rendering with Unposed Patch Transformer (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.