Research map / Detection and segmentation / Open-set and few-shot detection
End-to-End Open-Vocabulary Video Visual Relationship Detection Using Multi-Modal Prompting
Yongqi Wang, Xinxiao Wu, Shuo Yang, Jiebo Luo
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 |
|---|---|---|---|
| train trajectory detector and relationship classifier end-to-end (ablation) | separate training · VidVRD (novel split, SGDet) | mAP: 14.43 -> 15.04 | Improved |
| train end-to-end (ablation) | separate training · VidOR (novel split, SGDet) | mAP: 1.99 -> 2.45 | Improved |
| add relationship query with auxiliary relationship loss (ablation) | trajectory detector with Aoc · VidVRD (novel split, SGDet) | mAP: 14.58 -> 15.04 | Improved |
| add auxiliary object classifier (ablation) | trajectory detector with Rna · VidVRD (novel split, SGDet) | mAP_o: 27.33 -> 36.31 | Improved |
| add relationship query with auxiliary loss (ablation) | detector without Aoc · VidVRD (novel split, SGDet) | mAP_o: 26.17 -> 27.33 | Improved |
| add spatio-temporal visual prompting (ablation) | language prompting only · VidVRD (novel split, SGDet) | mAP: 9.72 -> 15.04 | Improved |
| add vision-guided language prompting (ablation) | visual prompting only · VidVRD (novel split, SGDet) | mAP: 11.46 -> 15.04 | Improved |
| add temporal modeling to spatial modeling (ablation) | spatio-temporal visual prompting · VidVRD (novel split, SGDet) | mAP: 12.14 -> 15.04 | Improved |
| add spatial modeling to temporal modeling (ablation) | spatio-temporal visual prompting · VidVRD (novel split, SGDet) | mAP: 11.58 -> 15.04 | Improved |
| vision-guided language prompting instead of manual templates (ablation) | language prompts · VidVRD (novel split, SGDet) | mAP: 11.46 -> 15.04 | Improved |
| mix continuous and conditional prompts instead of continuous only (ablation) | language prompts · VidVRD (novel split, SGDet) | mAP: 12.79 -> 15.04 | Improved |
| set box filtering threshold to 0.35 rather than 0.45 (ablation) | trajectory detector · VidVRD (novel split, SGDet) | mAP: 12.95 -> 15.04 | Improved |
Closest papers by meaning
- Multi-Modal Prompting for Open-Vocabulary Video Visual Relationship Detection (AAAI 2024 · similarity 0.95)
- Scene-Graph ViT: End-to-End Open-Vocabulary Visual Relationship Detection (ECCV 2024 · similarity 0.90)
- Compositional Prompt Tuning with Motion Cues for Open-vocabulary Video Relation Detection (ICLR 2023 · similarity 0.90)
- Towards Flexible Visual Relationship Segmentation (NeurIPS 2024 · similarity 0.87)
- Open-Vocabulary Video Relation Extraction (AAAI 2024 · similarity 0.87)
- OVTR: End-to-End Open-Vocabulary Multiple Object Tracking with Transformer (ICLR 2025 · similarity 0.86)
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.
