Research map / 3D vision / NeRF and radiance fields
Neural Radiance Field Codebooks
Matthew Wallingford, Aditya Kusupati, Alex Fang, Vivek Ramanujan, Aniruddha Kembhavi, Roozbeh Mottaghi, Ali Farhadi
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 |
|---|---|---|---|
| Use NRC rather than ObSuRF for unsupervised segmentation | NRC · ProcTHOR | ARI: .228 -> .295 | Improved |
| Use NRC rather than ObSuRF for unsupervised segmentation | NRC · NYU Depth | ARI: .141 -> .182 | Improved |
| Use NRC rather than ObSuRF for unsupervised segmentation | NRC · CLEVR-3D | FG-ARI: .978 -> .977 | Hurt |
| Use NRC instead of the strongest baseline representation | Object navigation policy · RoboTHOR object navigation test | Success Rate (%): 47.0 -> 50.1 | Improved |
| Use NRC instead of the strongest baseline representation | Object navigation policy · RoboTHOR object navigation test | SPL: .200 -> .239 | Improved |
| Use NRC rather than ObSuRF | Depth ordering · ProcTHOR test | Depth Order Acc. (%): 18.3 -> 23.8 | Improved |
| Add intra-code variation modeling (ablation) | NRC with fixed codebook size · ProcTHOR | ARI: .182 -> .284 | Improved |
| Learn codebook size (ablation) | NRC without variation module · ProcTHOR | ARI: .182 -> .197 | Improved |
| Learn codebook size in addition to intra-code variation (ablation) | NRC · ProcTHOR | ARI: .284 -> .295 | Improved |
| Vary fixed codebook size (ablation) | NRC without learned code count · ProcTHOR | ARI: 10: .247; 20: .256; 30: .266; 40: .279; 50: .286; 60: .293; 70: .288; 80: .295; 90: .279; 100: .273 | Mixed |
| Use NRC rather than ObSuRF on cluttered scenes | Unsupervised segmentation · NYU Depth (Hard) | ARI: .092 -> .139 | Improved |
| Use NRC rather than ObSuRF | Unsupervised segmentation · MatterPort3D | ARI: .178 -> .237 | Improved |
Closest papers by meaning
- GRF: Learning a General Radiance Field for 3D Representation and Rendering (ICCV 2021 · similarity 0.90)
- Learning Object-Compositional Neural Radiance Field for Editable Scene Rendering (ICCV 2021 · similarity 0.90)
- Learning Unified Decompositional and Compositional NeRF for Editable Novel View Synthesis (ICCV 2023 · similarity 0.89)
- Unsupervised Multi-View Object Segmentation Using Radiance Field Propagation (NeurIPS 2022 · similarity 0.89)
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ECCV 2020 · similarity 0.88)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · similarity 0.87)
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.
