Research map / 3D vision / 3D generation
ReScene4D: Temporally Consistent Semantic Instance Segmentation of Evolving Indoor 3D Scenes
Emily Steiner, Jianhao Zheng, Henry Howard-Jenkins, Chris Xie, Iro Armeni
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 cross-time contrastive loss (ablation) | ReScene4D (C) · 3RScan 4DSIS (Concerto) | t-mAP: 28.4 -> 34.1 | Improved |
| add spatio-temporal serialization (ablation) | ReScene4D (C) · 3RScan 4DSIS (Concerto) | t-mAP: 28.4 -> 32.9 | Improved |
| add spatio-temporal masking (ablation) | ReScene4D (C) · 3RScan 4DSIS (Concerto) | t-mAP: 28.4 -> 32.4 | Improved |
| add ST-serial on top of contrastive loss (ablation) | ReScene4D (C) · 3RScan 4DSIS (Concerto) | t-mAP: 34.1 -> 34.8 | Improved |
| add all three strategies versus contrastive plus serialization (ablation) | ReScene4D (C) · 3RScan 4DSIS (Concerto) | t-mAP: 34.8 -> 33.3 | Hurt |
| add contrastive loss (ablation) | ReScene4D (S) Sonata · 3RScan 4DSIS (Sonata) | t-mAP: 29.7 -> 25.7 | Hurt |
| add contrastive loss (ablation) | ReScene4D (M) Minkowski · 3RScan 4DSIS (Minkowski) | t-mAP: 32.0 -> 31.5 | No clear effect |
| use mixed spatial and temporal serialization instead of spatial only (ablation) | Concerto · 3RScan 4DSIS | t-mAP: 28.4 -> 32.9 | Improved |
| use mixed serialization instead of temporal only (ablation) | Concerto · 3RScan 4DSIS | t-mAP: 32.0 -> 32.9 | Improved |
| use 4D scene with contrastive loss instead of 3D without (ablation) | Concerto · 3RScan SIS | mAP: 42.3 -> 47.3 | Improved |
| replace Mask3D with ReScene4D (C) for per-stage 3DSIS | ReScene4D (C) · 3RScan per-stage 3DSIS | mAP: 46.4 -> 47.8 | Improved |
| use geometric instead of semantic matching for Mask3D | Mask3D · 3RScan per-stage 3DSIS | mAP: 46.4 -> 21.9 | Hurt |
Closest papers by meaning
- Online Segment Any 3D Thing as Instance Tracking (NeurIPS 2025 · similarity 0.88)
- Living Scenes: Multi-object Relocalization and Reconstruction in Changing 3D Environments (CVPR 2024 · similarity 0.87)
- IGGT4D: Streaming 4D Instance-Grounded Geometry Transformer (NeurIPS 2026 · similarity 0.86)
- Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidance (CVPR 2024 · similarity 0.86)
- Motion4D: Learning 3D-Consistent Motion and Semantics for 4D Scene Understanding (NeurIPS 2025 · similarity 0.86)
- UnScene3D: Unsupervised 3D Instance Segmentation for Indoor Scenes (CVPR 2024 · similarity 0.85)
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.
