Research map / 3D vision / NeRF and radiance fields
SPIn-NeRF: Multiview Segmentation and Perceptual Inpainting With Neural Radiance Fields
Read the paperCheck 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 two-stage multiview segmentation instead of video segmentation alone | SPIn-NeRF multiview segmentation · Multiview segmentation | IoU: 88.34 -> 91.66 | Helped |
| Use perceptual NeRF inpainting rather than LaMa (2D) | SPIn-NeRF · 3D inpainting (human masks) | LPIPS: 0.5369 -> 0.4654 | Helped |
| Use perceptual NeRF inpainting rather than LaMa (2D) | SPIn-NeRF · 3D inpainting (human masks) | FID: 174.61 -> 156.64 | Helped |
| Use predicted masks instead of human-annotated masks | SPIn-NeRF inpainting · 3D inpainting | LPIPS: 0.4654 -> 0.4662 | Hurt |
| Add depth geometry guidance (ablation) | SPIn-NeRF inpainting · 3D inpainting (human masks) | LPIPS: 0.4952 -> 0.4654 | Helped |
| Refine both RGB and depth (ablation) | SPIn-NeRF inpainting · 3D inpainting (human masks) | FID: 156.64 -> 163.79 | Hurt |
| Refine RGB without refining depth (ablation) | SPIn-NeRF inpainting · 3D inpainting (human masks) | FID: 156.64 -> 147.31 | Helped |
| Use two training stages instead of one (ablation) | SPIn-NeRF multiview segmentation · Multiview segmentation | IoU: 90.96 -> 91.66 | Helped |
| Use three training stages instead of two (ablation) | SPIn-NeRF multiview segmentation · Multiview segmentation | IoU: 91.66 -> 91.53 | No clear effect |
| Reduce input views from 60 to 10 (ablation) | SPIn-NeRF inpainting · 3D inpainting | LPIPS: 0.4654 -> 0.4726 | Hurt |
| Dilate masks by 45 iterations instead of 5 (ablation) | SPIn-NeRF inpainting · 3D inpainting | LPIPS: 0.4654 -> 0.6102 | Hurt |
| Use undilated masks instead of masks dilated by 5 iterations (ablation) | SPIn-NeRF inpainting · 3D inpainting | LPIPS: 0.4654 -> 0.4904 | Hurt |
Closest papers by meaning
- MVIP-NeRF: Multi-view 3D Inpainting on NeRF Scenes via Diffusion Prior (CVPR 2024 · similarity 0.91)
- InNeRF360: Text-Guided 3D-Consistent Object Inpainting on 360-degree Neural Radiance Fields (CVPR 2024 · similarity 0.90)
- SANeRF-HQ: Segment Anything for NeRF in High Quality (CVPR 2024 · similarity 0.89)
- NeRF-HuGS: Improved Neural Radiance Fields in Non-static Scenes Using Heuristics-Guided Segmentation (CVPR 2024 · similarity 0.89)
- Language-driven Object Fusion into Neural Radiance Fields with Pose-Conditioned Dataset Updates (CVPR 2024 · similarity 0.89)
- NeRFDeformer: NeRF Transformation from a Single View via 3D Scene Flows (CVPR 2024 · 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.