Research map / Video / Video-language understanding
Open-Book Video Captioning With Retrieve-Copy-Generate Network
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 |
|---|---|---|---|
| Add retrieved sentence guidance through a fixed retriever | Hierarchical-LSTM captioning on MSR-VTT · MSR-VTT test | CIDEr: 49.8 -> 52.3 | Helped |
| Add retrieved sentence guidance with joint retriever training | Hierarchical-LSTM captioning on VATEX · VATEX test | CIDEr: 49.2 -> 57.5 | Helped |
| Add ResNet appearance and I3D motion features to the retriever (ablation) | RCG with a fixed VATEX retriever · VATEX test | CIDEr: 49.2 -> 56.8 | Helped |
| Replace the random retriever with a ResNet-only retriever (ablation) | RCG with a fixed VATEX retriever · VATEX test | CIDEr: 49.1 -> 51.5 | Helped |
| Replace the random retriever with an I3D-only retriever (ablation) | RCG with a fixed VATEX retriever · VATEX test | CIDEr: 49.1 -> 54.7 | Helped |
| Jointly train the retriever instead of keeping it fixed, using 3 retrievals during training (ablation) | RCG on VATEX · VATEX test | CIDEr: 56.8 -> 57.5 | Helped |
| Train with 3 retrieved sentences instead of 1, with a fixed retriever (ablation) | RCG on VATEX · VATEX test | CIDEr: 55.3 -> 56.8 | Helped |
| Train with 5 or 10 retrieved sentences instead of 3, with a fixed retriever (ablation) | RCG on VATEX · VATEX test | CIDEr: 56.8 -> 56.3; 56.8 -> 56.4 | Hurt |
| Increase the number of retrieved sentences at inference from 1 to 30 (ablation) | RCG trained with 3 retrieved sentences · VATEX test | CIDEr: 50.5 -> 57.5 | Helped |
| Expand the retrieval corpus from +0.1% to +100% (ablation) | RCG on VATEX · VATEX test | CIDEr: 39.9 -> 56.8 | Helped |
| Use an Oracle retrieval corpus containing test-set sentences (ablation) | RCG on VATEX · VATEX test | CIDEr: 56.8 -> 58.9 | Helped |
Closest papers by meaning
- Narrating the Video: Boosting Text-Video Retrieval via Comprehensive Utilization of Frame-Level Captions (CVPR 2025 · similarity 0.86)
- Distilling Vision-Language Models on Millions of Videos (CVPR 2024 · similarity 0.86)
- VideoCon: Robust Video-Language Alignment via Contrast Captions (CVPR 2024 · similarity 0.86)
- Panda-70M: Captioning 70M Videos with Multiple Cross-Modality Teachers (CVPR 2024 · similarity 0.85)
- OmniViD: A Generative Framework for Universal Video Understanding (CVPR 2024 · similarity 0.85)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · 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.