Research map / 3D vision / 3D generation
GNeSF: Generalizable Neural Semantic Fields
Hanlin Chen, Chen Li, Mengqi Guo, Zhiwen Yan, Gim Hee Lee
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 soft voting of 2D semantic observations instead of a direct semantic head | GNeSF · ScanNet val, novel scene | mIoU: 39.1 -> 71.6 | Improved |
| Use GNeSF instead of Semantic-Ray | ScanNet val, novel scene · ScanNet val, novel scene | mIoU: 56.0 -> 71.6 | Improved |
| Use GNeSF instead of Mask2Former with rendered images | ScanNet val, novel scene · ScanNet val, novel scene | mIoU: 64.6 -> 71.6 | Improved |
| Use GNeSF instead of Semantic-NeRF | ScanNet val, novel view · ScanNet val, novel view | mIoU: 96.8 -> 93.3 | Hurt |
| Use GNeSF instead of NeuralRecon | ScanNet test 3D semantic segmentation · ScanNet test, 3D semantic segmentation | mIoU: 51.5 -> 55.1 | Improved |
| Apply the ScanNet-trained GNeSF model to Replica | GNeSF · Replica, 3D semantic segmentation | 3D mIoU: 26.2 -> 43.8 | Improved |
| Use mesh geometry (ablation) | GNeSF · ScanNet val, 3D semantic segmentation | mIoU: 58.8 -> 60.4 | Improved |
| Add semantic voting instead of direct point-semantic prediction (ablation) | GNeSF · ScanNet 3D semantic segmentation | mIoU: 37.5 (+0) -> 57.3 (+19.8) | Improved |
| Represent semantics as probabilities rather than logits (ablation) | GNeSF soft voting · ScanNet 3D semantic segmentation | mIoU: 57.3 -> 57.9 (+0.6) | Improved |
| Add the visibility module to filter occluded views (ablation) | GNeSF probability voting · ScanNet 3D semantic segmentation | mIoU: 57.9 -> 58.8 (+0.9) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- NeuralRecon — NeuralRecon: Real-Time Coherent 3D Reconstruction From Monocular Video (CVPR 2021) · better on 1 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
- GOV-NeSF: Generalizable Open-Vocabulary Neural Semantic Fields (CVPR 2024 · similarity 0.91)
- GSNeRF: Generalizable Semantic Neural Radiance Fields with Enhanced 3D Scene Understanding (CVPR 2024 · similarity 0.88)
- Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance (AAAI 2025 · similarity 0.87)
- A Unified Framework for 3D Scene Understanding (NeurIPS 2024 · similarity 0.87)
- Large Spatial Model: End-to-end Unposed Images to Semantic 3D (NeurIPS 2024 · similarity 0.87)
- Segment Any 3D Object with Language (ICLR 2025 · 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.
