Research map / 3D vision / NeRF and radiance fields
Learning Heterogeneous Mixture of Scene Experts for Large-Scale Neural Radiance Fields
Zhenxing Mi, Ping Yin, Xiao Xue, 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 heterogeneous hash experts instead of homogeneous (ablation) | Switch-NeRF++ · UrbanBIS | PSNR: 20.42 -> 20.76 | Improved |
| add heterogeneous hash experts instead of homogeneous (ablation) | Switch-NeRF++ · Block-NeRF dataset | PSNR: 23.90 -> 24.11 | Improved |
| use hash-based gating instead of MLP gating (ablation) | Switch-NeRF++ · Residence | PSNR: 22.03 -> 23.25 | Improved |
| use hash-based gating instead of MLP gating (ablation) | Switch-NeRF++ · Sci-Art | PSNR: 24.00 -> 26.17 | Improved |
| use 8 experts instead of 2 (ablation) | Switch-NeRF++ · Residence | PSNR: 22.62 -> 23.25 | Improved |
| use Top-2 instead of Top-1 gating (ablation) | Switch-NeRF++ · Residence | PSNR: 23.25 -> 23.33 | No clear effect |
| use hash dispatching instead of MoE dispatching (ablation) | Switch-NeRF++ · Building training time | hours: 15.2h -> 10.7h | Improved |
| add balance loss L_b to gating (ablation) | Switch-NeRF · Building | PSNR: 15.32 -> 21.54 | Improved |
| use full dispatch instead of uniform dispatch Cf=2.0 (ablation) | Switch-NeRF · Building | PSNR: 17.82 -> 21.54 | Improved |
| add LayerNorm to MLP gating (ablation) | Switch-NeRF · Building | PSNR: 21.39 -> 21.54 | Improved |
| use two gating operations instead of one (ablation) | Switch-NeRF · Building | PSNR: 21.54 -> 21.55 | No clear effect |
| use Top-2 and larger capacity in training (ablation) | Switch-NeRF · Building | PSNR: 21.54 -> 21.77 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- VastGaussian — VastGaussian: Vast 3D Gaussians for Large Scene Reconstruction (CVPR 2024) · better on 0 of 1 shared result
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
- Switch-NeRF: Learning Scene Decomposition with Mixture of Experts for Large-scale Neural Radiance Fields (ICLR 2023 · similarity 0.94)
- NeRFLight: Fast and Light Neural Radiance Fields Using a Shared Feature Grid (CVPR 2023 · similarity 0.91)
- NeRFusion: Fusing Radiance Fields for Large-Scale Scene Reconstruction (CVPR 2022 · similarity 0.91)
- Block-NeRF: Scalable Large Scene Neural View Synthesis (CVPR 2022 · similarity 0.91)
- Rad-NeRF: Ray-decoupled Training of Neural Radiance Field (NeurIPS 2024 · similarity 0.90)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · 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.
