Research map / Video / Video-language understanding
Collaborative Static and Dynamic Vision-Language Streams for Spatio-Temporal Video Grounding
Zihang Lin, Chaolei Tan, Jian-Fang Hu, Zhi Jin, Tiancai Ye, Wei-Shi Zheng
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 |
|---|---|---|---|
| Use collaborative static and dynamic vision-language streams | STVG grounding model · HCSTVG-v1 test | m vIoU: 32.4 -> 36.9 | Improved |
| Use collaborative static and dynamic vision-language streams | STVG grounding model · HCSTVG-v2 val | m vIoU: 36.4 -> 38.7 | Improved |
| Use collaborative static and dynamic vision-language streams | STVG grounding model · VidSTG declarative test | m vIoU: 30.4 -> 33.7 | Improved |
| Use collaborative static and dynamic vision-language streams | STVG grounding model · VidSTG interrogative test | m vIoU: 25.7 -> 28.5 | Improved |
| Add the cross-stream collaboration block (ablation) | STVG grounding model · HCSTVG-v2 val | m vIoU: 36.4 -> 38.7 | Improved |
| Add static-to-dynamic information transmission (ablation) | STVG grounding model · HCSTVG-v2 val | m tIoU: 56.1 -> 57.4 | Improved |
| Add dynamic-to-static information transmission (ablation) | STVG grounding model · HCSTVG-v2 val | m sIoU: 64.9 -> 65.7 | Improved |
| Retain spatial information in the dynamic-stream video features (ablation) | Dynamic VL stream · HCSTVG-v2 ablation | m tIoU: 54.7 -> 56.1 | Improved |
| Remove spatial mean pooling before text-to-visual cross-attention (ablation) | Dynamic VL stream · HCSTVG-v2 ablation | m tIoU: 56.1 -> 56.2 | No clear effect |
| Use asymmetric cross-stream collaboration instead of sum fusion (ablation) | STVG grounding model · HCSTVG-v2 val | m vIoU: 36.6 -> 38.7 | Improved |
| Use asymmetric cross-stream collaboration instead of FC + sum fusion (ablation) | STVG grounding model · HCSTVG-v2 val | m vIoU: 36.8 -> 38.7 | Improved |
| Use asymmetric cross-stream collaboration instead of concat + FC fusion (ablation) | STVG grounding model · HCSTVG-v2 val | m vIoU: 37.1 -> 38.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TubeDETR — TubeDETR: Spatio-Temporal Video Grounding With Transformers (CVPR 2022) · better on 12 of 12 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- STPro: Spatial and Temporal Progressive Learning for Weakly Supervised Spatio-Temporal Grounding (CVPR 2025 · similarity 0.90)
- Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding (NeurIPS 2025 · similarity 0.89)
- Context-Guided Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.89)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.89)
- Motion-Grounded Video Reasoning: Understanding and Perceiving Motion at Pixel Level (CVPR 2025 · similarity 0.88)
- Time-R1: Post-Training Large Vision Language Model for Temporal Video Grounding (NeurIPS 2025 · 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.
