Research map / Point clouds / Point cloud classification
Scan2Cap: Context-Aware Dense Captioning in RGB-D Scans
Check 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 |
|---|---|---|---|
| Caption directly from 3D detections rather than 2D detections projected into 3D | Dense captioning on ScanRefer · ScanRefer val | CiDEr@0.5IoU: 11.47 -> 39.08 | Helped |
| Add context-aware attention captioning to basic GRU captioning (ablation) | VoteNet captioning · ScanRefer val | CiDEr@0.5IoU: 34.31 -> 36.15 | Helped |
| Add a relational graph to context-aware attention captioning (ablation) | VoteNet + CAC · ScanRefer val | CiDEr@0.5IoU: 36.15 -> 39.08 | Helped |
| Fine-tune VoteNet end-to-end with description-generation and relative-orientation losses (ablation) | Scan2Cap · ScanRefer val | CiDEr@0.5IoU: 33.87 -> 39.08 | Helped |
| Add multiview image features to xyz input (ablation) | Scan2Cap · ScanRefer val | CiDEr@0.5IoU: 32.94 -> 36.89 | Helped |
| Add normals to xyz+multiview input (ablation) | Scan2Cap · ScanRefer val | CiDEr@0.5IoU: 36.89 -> 39.08 | Helped |
| Caption from 3D object boxes rather than 2D object boxes | Oracle captioning with GT boxes · ScanRefer val, GT boxes | CiDEr@0.5IoU: 58.44 -> 67.95 | Helped |
| Use RGB features with xyz coordinates (ablation) | Scan2Cap · ScanRefer val | CiDEr@0.5IoU: 32.94 -> 33.41 | Helped |
| Add normals to xyz+rgb input (ablation) | Scan2Cap · ScanRefer val | CiDEr@0.5IoU: 33.41 -> 35.20 | Helped |
Closest papers by meaning
- Dense Captioning With Joint Inference and Visual Context (CVPR 2017 · similarity 0.91)
- Dense Video Object Captioning from Disjoint Supervision (ICLR 2025 · similarity 0.88)
- Benchmarking Large Vision-Language Models via Directed Scene Graph for Comprehensive Image Captioning (CVPR 2025 · similarity 0.87)
- A Picture is Worth More Than 77 Text Tokens: Evaluating CLIP-Style Models on Dense Captions (CVPR 2024 · similarity 0.86)
- ResCLIP: Residual Attention for Training-free Dense Vision-language Inference (CVPR 2025 · similarity 0.85)
- FlexCap: Describe Anything in Images in Controllable Detail (NeurIPS 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.