Research map / 3D vision / Gaussian splatting
N2F2: Hierarchical Scene Understanding with Nested Neural Feature Fields
Yash Bhalgat, Iro Laina, Joao F Henriques, Andrew Zisserman, Andrea Vedaldi
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 composite embedding instead of explicit scale selection (ablation) | N2F2 · expanded LERF (4 scenes) | mIoU: 53.7 -> 54.4 | Improved |
| use composite embedding instead of explicit scale selection (ablation) | N2F2 · expanded LERF (4 scenes) | localization accuracy: 87.0 -> 87.9 | Improved |
| use all dimensions for scale supervision (k=1 vs k=170) (ablation) | N2F2 · LERF teatime | mIoU: 66.4 -> 69.2 | Improved |
| use composite embedding to compute one relevancy map per query | N2F2 · querying time | speed: 1.7x faster than LangSplat, 26x faster than LERF | Improved |
| use a weaker segmenter (Detic instead of SAM) (ablation) | N2F2 · LERF teatime | mIoU: 69.2 -> 66.2 | Hurt |
| remove nested feature design (ablation) | N2F2 · LERF | mIoU: -2.0 mIoU, -2.5% localization when removed | Improved |
| use pure-MLP backbone instead of TriPlane+MLP (ablation) | N2F2 · LERF | convergence speed: 2.3x more memory-efficient but 20x slower convergence | Mixed |
| use N2F2 instead of LangSplat on simple 3D-OVS scenes | N2F2 · 3D-OVS | mIoU: 93.4 -> 93.9 | No clear effect |
| use N2F2 on compound queries instead of LangSplat | N2F2 · expanded LERF ramen compound queries | mIoU: 35.1 -> 42.3 | Improved |
Later papers that compared with this method
2 later accepted papers report N2F2 as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Segment Any 3D Gaussians (AAAI 2025 · better on 1 of 1 shared result)
- OpenHype: Hyperbolic Embeddings for Hierarchical Open-Vocabulary Radiance Fields (NeurIPS 2025 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- LangSplat — LangSplat: 3D Language Gaussian Splatting (CVPR 2024) · better on 5 of 5 shared results
- LERF — LERF: Language Embedded Radiance Fields (ICCV 2023) · better on 3 of 3 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
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 2024 · similarity 0.86)
- DistillNeRF: Perceiving 3D Scenes from Single-Glance Images by Distilling Neural Fields and Foundation Model Features (NeurIPS 2024 · similarity 0.86)
- GOV-NeSF: Generalizable Open-Vocabulary Neural Semantic Fields (CVPR 2024 · similarity 0.86)
- Lexicon3D: Probing Visual Foundation Models for Complex 3D Scene Understanding (NeurIPS 2024 · similarity 0.86)
- How to Train Neural Field Representations: A Comprehensive Study and Benchmark (CVPR 2024 · similarity 0.86)
- Masked Scene Modeling: Narrowing the Gap Between Supervised and Self-Supervised Learning in 3D Scene Understanding (CVPR 2025 · similarity 0.86)
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.
