Research map / 3D vision / 3D generation
Holo-Captioning: A Comprehensive Textual View of 3D Scenes
Kun-Yu Lin, Chengke Bu, Zhenguo Li, Kai Han
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 decoupling of relation prediction (ablation) | directly tuned SpatialLM · HoloScan validation | Overall HoloScore: 58.64 -> 71.01 | Improved |
| add instance-aware attribute and relation prediction (ablation) | HoloScribe · HoloScan validation | Overall HoloScore: 71.01 -> 109.98 | Improved |
| add instance-aware attribute and relation prediction (ablation) | HoloScribe · HoloScan validation | Attribute F-score: 6.74 -> 22.41 | Improved |
| add anchor-aware instance linking (ablation) | HoloScribe · HoloScan validation | Relation F-score: 7.92 -> 11.08 | Improved |
| add anchor-aware instance linking (ablation) | HoloScribe · HoloScan validation | Overall HoloScore: 108.27 -> 116.36 | Improved |
| add anchor-aware instance linking (ablation) | HoloScribe · HoloScan validation | Tagging F-score: 58.67 -> 60.43 | Improved |
| add anchor-aware instance linking (ablation) | HoloScribe · HoloScan validation | Location F-score: 17.94 -> 22.33 | Improved |
| add anchor-aware instance linking (ablation) | HoloScribe · HoloScan validation | Attribute F-score: 22.46 -> 22.52 | No clear effect |
| HoloScribe vs SpatialLM-Tuned, same Qwen2.5-0.5B LLM | HoloScribe · HoloScan test | Overall HoloScore: 69.63 -> 117.04 | Improved |
| HoloScribe vs SpatialLM zero-shot on MultiScan | HoloScribe · MultiScan detection (zero-shot) | Tagging F-score: 20.2 -> 53.5 | Improved |
| HoloScribe vs SpatialLM zero-shot on MultiScan | HoloScribe · MultiScan detection (zero-shot) | Location F-score: 9.6 -> 20.2 | Improved |
| HoloScribe vs LL3DA-Tuned on relation, test set | HoloScribe · HoloScan test | Relation F-score: 3.97 -> 8.74 | Improved |
Closest papers by meaning
- OpenLex3D: A Tiered Benchmark for Open-Vocabulary 3D Scene Representations (NeurIPS 2025 · similarity 0.86)
- The Scene Language: Representing Scenes with Programs, Words, and Embeddings (CVPR 2025 · similarity 0.85)
- SceneForge: Enhancing 3D-text alignment with Structured Scene Compositions (NeurIPS 2025 · similarity 0.85)
- MARVEL-40M+: Multi-Level Visual Elaboration for High-Fidelity Text-to-3D Content Creation (CVPR 2025 · similarity 0.85)
- MMScan: A Multi-Modal 3D Scene Dataset with Hierarchical Grounded Language Annotations (NeurIPS 2024 · similarity 0.85)
- GraphDreamer: Compositional 3D Scene Synthesis from Scene Graphs (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.
