Research map / Video / Video-language understanding
ScanFocus: A Coarse-to-Fine Framework for Spatio-Temporal Video Grounding
Chen Kai, Ming Dai, Wenxuan Cheng, Wankou Yang
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 dense boundary sampling with refine decoders (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 50.9 -> 52.4 | Improved |
| add dense boundary sampling without SGTA (ablation) | ScanFocus · HC-STVGv1 | vIoU@0.5: 38.0 -> 37.9 | No clear effect |
| add SGTA on top of dense sampling (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 52.4 -> 53.7 | Improved |
| add temporal attention in SGTA (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 52.8 -> 53.7 | Improved |
| add semantic guide in SGTA (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 53.1 -> 53.7 | Improved |
| increase refine window size beyond 8 (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 53.7 -> 53.5 (N_w 8 -> 12) | Hurt |
| increase refine window size from 4 to 8 (ablation) | ScanFocus · HC-STVGv1 | m_tIoU: 52.7 -> 53.7 | Improved |
| use deformable semantic-motion fusion instead of standard self-attention (ablation) | ScanFocus · HC-STVGv1 | vIoU@0.5: 36.5 -> 39.5 | Improved |
| use deformable fusion instead of standard self-attention (ablation) | ScanFocus · HC-STVGv1 | GFLOPs: 516 -> 260 | Improved |
| ScanFocus vs prior best TA-STVG on HC-STVGv1 | ScanFocus · HC-STVGv1 | vIoU@0.5: 36.8 -> 42.2 | Improved |
| ScanFocus vs TA-STVG on HC-STVGv2 | ScanFocus · HC-STVGv2 | m_tIoU: 60.4 -> 62.4 | Improved |
| ScanFocus vs TA-STVG on VidSTG declarative | ScanFocus · VidSTG declarative | vIoU@0.5: 33.5 -> 36.0 | Improved |
Closest papers by meaning
- Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding (NeurIPS 2025 · similarity 0.91)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.91)
- Context-Guided Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.90)
- STPro: Spatial and Temporal Progressive Learning for Weakly Supervised Spatio-Temporal Grounding (CVPR 2025 · similarity 0.90)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 2025 · similarity 0.90)
- Contextual Self-paced Learning for Weakly Supervised Spatio-Temporal Video Grounding (ICLR 2025 · 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.
