Research map / 3D vision / NeRF and radiance fields
Segment Anything in 3D with NeRFs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 confidence decay term to self-prompting (vs plain 2D NMS) (ablation) | SA3D · NVOS mean | mIoU (%): 87.9 -> 90.3 | Helped |
| add confidence decay term to self-prompting (ablation) | SA3D · NVOS Trex | IoU (%): 64.0 -> 82.0 | Helped |
| alternate mask inverse rendering and self-prompting across views (vs single view) (ablation) | SA3D · SPIn-NeRF mean | IoU (%): 74.6 -> 92.4 | Helped |
| alternate mask inverse rendering and self-prompting across views (vs single view) (ablation) | SA3D · Replica mean | mIoU (%): 59.8 -> 83.0 | Helped |
| use more views for 3D mask generation on a 360 scene (ablation) | SA3D · Lego (360) | IoU (%): 84.5 -> 92.2 | Helped |
| use more views on a forward-facing scene (ablation) | SA3D · Fortress | IoU (%): 97.8 -> 98.3 | No clear effect |
| IoU rejection threshold tau varied from 0.3 to 0.7 (ablation) | SA3D · Replica office0 | mIoU: 74.9 (0.3), 84.4 (0.5), 82.0 (0.7) | Mixed |
| negative term coefficient lambda varied from 0.05 to 0.5 (ablation) | SA3D · Replica office0 | mIoU: 79.1 (0.05), 84.9 (0.3), 83.3 (0.5) | Mixed |
| replace SAM with other prompt-based 2D segmenters (SEEM, SimpleClick, RITM, FocalClick) (ablation) | SA3D · NVOS | mIoU (%): 86.0 / 87.7 / 81.2 / 88.9 vs 90.3 with SAM | Mixed |
| SA3D vs ISRF | NVOS · NVOS | mIoU (%): 83.8 -> 90.3 | Helped |
Closest papers by meaning
- SANeRF-HQ: Segment Anything for NeRF in High Quality (CVPR 2024 · similarity 0.93)
- NTO3D: Neural Target Object 3D Reconstruction with Segment Anything (CVPR 2024 · similarity 0.88)
- SAM2Object: Consolidating View Consistency via SAM2 for Zero-Shot 3D Instance Segmentation (CVPR 2025 · similarity 0.88)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · similarity 0.88)
- EmbodiedSAM: Online Segment Any 3D Thing in Real Time (ICLR 2025 · similarity 0.88)
- SA3DIP: Segment Any 3D Instance with Potential 3D Priors (NeurIPS 2024 · similarity 0.88)
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.