Research map / 3D vision / NeRF and radiance fields
Empowering Sparse-Input Neural Radiance Fields with Dual-Level Semantic Guidance from Dense Novel Views
Yingji Zhong, Kaichen Zhou, Zhihao Li, Lanqing Hong, Zhenguo Li, Dan Xu
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 supervision-level guidance (ablation) | teacher NeRF baseline · ScanNet++ | PSNR: 21.17 -> 21.69 | Improved |
| add Bi-Directional Verification (ablation) | supervision-level guidance · ScanNet++ | LPIPS: 0.409 -> 0.387 | Improved |
| use rendered semantic labels from novel views rather than source views only (ablation) | supervision-level guidance · ScanNet++ | PSNR: 20.49 -> 21.69 | Improved |
| add feature-level guidance codebook (ablation) | S3NeRF · ScanNet++ | PSNR: 21.69 -> 22.21 | Improved |
| add semantic loss to codebook (ablation) | S3NeRF · ScanNet++ | LPIPS: 0.376 -> 0.364 | Improved |
| use learnable semantic-aware codebook instead of pre-trained (ablation) | S3NeRF · ScanNet++ | PSNR: 20.28 -> 22.21 | Improved |
| add rendered RGB as novel-view supervision (ablation) | baseline · Replica | LPIPS: 0.318 -> 0.320 | Hurt |
| filter rendered RGB with validity map (ablation) | rendered RGB supervision · Replica | LPIPS: 0.320 -> 0.312 | Improved |
| use filtered rendered semantics instead of rendered RGB (ablation) | baseline · Replica | PSNR: 21.72 -> 22.34 | Improved |
| add warped RGB (filtered) (ablation) | baseline · Replica | PSNR: 21.37 -> 21.36 | No clear effect |
| vary codebook size K from 8 to 128 (ablation) | S3NeRF · ScanNet++ (scene a29cccc784) | PSNR: 21.87 vs 22.02 vs 21.89 | No clear effect |
| use codebook at MLP layer idx 3 (ablation) | S3NeRF · ScanNet++ (scene a29cccc784) | PSNR: 21.81 -> 17.50 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- FreeNeRF — FreeNeRF: Improving Few-Shot Neural Rendering With Free Frequency Regularization (CVPR 2023) · better on 4 of 4 shared results
- DNGaussian — DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization (CVPR 2024) · better on 2 of 2 shared results
- SparseNeRF — SparseNeRF: Distilling Depth Ranking for Few-shot Novel View Synthesis (ICCV 2023) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Semantically-aware Neural Radiance Fields for Visual Scene Understanding: A Comprehensive Review (IJCV 2026 · similarity 0.91)
- NeRF-VPT: Learning Novel View Representations with Neural Radiance Fields via View Prompt Tuning (AAAI 2024 · similarity 0.91)
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ECCV 2020 · similarity 0.91)
- Few-Shot Neural Radiance Fields under Unconstrained Illumination (AAAI 2024 · similarity 0.90)
- GSNeRF: Generalizable Semantic Neural Radiance Fields with Enhanced 3D Scene Understanding (CVPR 2024 · similarity 0.90)
- Putting NeRF on a Diet: Semantically Consistent Few-Shot View Synthesis (ICCV 2021 · similarity 0.90)
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.
