Research map / Video / Video-language understanding
Local-Global Video-Text Interactions for Temporal Grounding
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 |
|---|---|---|---|
| Use LGI rather than the strongest baseline | Temporal grounding · Charades-STA | R@0.5: 52.02 -> 59.46 | Helped |
| Use LGI rather than the strongest baseline | Temporal grounding · ActivityNet Captions | R@0.5: 36.90 -> 41.51 | Helped |
| Add distinct query attention loss (ablation) | LGI · Charades-STA | R@0.5: 58.28 -> 59.46 | Helped |
| Add temporal attention guidance loss (ablation) | LGI · Charades-STA | R@0.5: 47.12 -> 59.46 | Helped |
| Extract phrase-level query features instead of using only a sentence-level feature (ablation) | LGI · Charades-STA | R@0.5: 57.34 -> 59.46 | Helped |
| Add local context modeling with ResBlock (ablation) | LGI-SQAN · Charades-STA | R@0.5: 40.86 -> 50.54 | Helped |
| Add global context modeling with NLBlock (ablation) | LGI-SQAN · Charades-STA | R@0.5: 40.86 -> 48.12 | Helped |
| Combine local ResBlock and global NLBlock context modeling (ablation) | LGI-SQAN · Charades-STA | R@0.5: 40.86 -> 57.34 | Helped |
| Perform segment-level modality fusion before context modeling (ablation) | LGI-SQAN · Charades-STA | R@0.5: 46.96 -> 57.34 | Helped |
| Use Hadamard product rather than addition or concatenation for modality fusion (ablation) | LGI-SQAN · Charades-STA | R@0.5: 46.75, 48.15 -> 57.34 | Helped |
| Add video position embedding (ablation) | LGI-SQAN · Charades-STA | R@0.5: 45.70 -> 57.34 | Helped |
| Increase the number of extracted semantic phrases beyond the best setting (ablation) | SQAN · Charades-STA and ActivityNet Captions | Recall at tIoU thresholds | Hurt |
Closest papers by meaning
- Empower Words: DualGround for Structured Phrase and Sentence-Level Temporal Grounding (NeurIPS 2025 · similarity 0.91)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.90)
- Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network (AAAI 2025 · similarity 0.90)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.90)
- Learning Multi-Scale Video-Text Correspondence for Weakly Supervised Temporal Article Gronding (AAAI 2024 · similarity 0.90)
- Comprehensive Visual Grounding for Video Description (AAAI 2024 · similarity 0.89)
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.