Research map / Video / Video-language understanding
Visual Relation Grounding in Videos
Junbin Xiao, Xindi Shang, Xun Yang, Sheng Tang, Tat-Seng Chua
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 message passing between subject and object (ablation) | vRGV · ImageNet-VidVRD | Acc_R (%): 17.63 -> 24.58 | Improved |
| add clip-level attention (hierarchical structure) (ablation) | vRGV · ImageNet-VidVRD | Acc_R (%): 20.58 -> 24.58 | Improved |
| add temporal attention unit (ablation) | vRGV · ImageNet-VidVRD | Acc_R (%): 7.06 -> 24.58 | Improved |
| vary temporal attention threshold from 0.01 to 0.05 (ablation) | vRGV · ImageNet-VidVRD | Acc_R (%): 24.51, 24.76, 24.43, 24.58, 22.94 | No clear effect |
| link regions with explicit object locations (post-linking) instead of greedy attention linking (ablation) | vRGV · ImageNet-VidVRD | Acc_R (%): 20.54 -> 24.58 | Improved |
| link regions with explicit object locations (post-linking) in co-occurrence baseline (ablation) | Co-occur · ImageNet-VidVRD | Acc_R (%): 12.80 -> 16.48 | Improved |
| use vRGV instead of T-Rank V1 | grounding · ImageNet-VidVRD | Acc_R (%): 3.99 -> 24.58 | Improved |
| use vRGV instead of Co-occur | grounding · ImageNet-VidVRD | Acc_R (%): 16.48 -> 24.58 | Improved |
| concatenate trajectory pairs (T-Rank V2 vs V1) | T-Rank · ImageNet-VidVRD | Acc_O (%): 10.68 -> 7.35 | Hurt |
| use vRGV instead of Co-occur in zero-shot | grounding · ImageNet-VidVRD zero-shot | Acc_R (%): 7.38 -> 10.27 | Improved |
| move from static to dynamic relations | Co-occur · ImageNet-VidVRD | Acc_R (%): 21.4 -> 13.81 | Hurt |
Closest papers by meaning
- Where Does It Exist: Spatio-Temporal Video Grounding for Multi-Form Sentences (CVPR 2020 · similarity 0.92)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.89)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.88)
- Comprehensive Visual Grounding for Video Description (AAAI 2024 · similarity 0.88)
- Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding (NeurIPS 2025 · similarity 0.88)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.88)
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.
