Research map / Video / Video-language understanding
STVGBert: A Visual-Linguistic Transformer Based Framework for Spatio-Temporal Video Grounding
Rui Su, Qian Yu, Dong Xu
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 the temporal grounding (TVG) branch (ablation) | STVGBert · VidSTG declarative (ablation, no temporal GT) | m_vIoU (%): 18.94 -> 23.97 | Improved |
| add the spatial grounding (SVG) branch (ablation) | STVGBert · VidSTG declarative (ablation, no temporal GT) | m_vIoU (%): 7.21 -> 23.97 | Improved |
| use one-stage joint prediction instead of two-stage (ablation) | STVGBert · VidSTG declarative (ablation, no temporal GT) | m_vIoU (%): 22.51 -> 23.97 | Improved |
| use one-stage joint prediction instead of two-stage (ablation) | STVGBert · VidSTG interrogative (ablation, no temporal GT) | m_vIoU (%): 21.05 -> 22.51 | Improved |
| use ST-ViLBERT instead of ViLBERT (ablation) | STVGBert · VidSTG declarative (ablation, no temporal GT) | m_vIoU (%): 21.27 -> 23.97 | Improved |
| use ST-ViLBERT instead of ViLBERT (with temporal GT) (ablation) | STVGBert · VidSTG declarative (ablation, with temporal GT) | m_vIoU (%): 41.75 -> 47.25 | Improved |
| use ST-ViLBERT instead of ViLBERT (with temporal GT) (ablation) | STVGBert · VidSTG interrogative (ablation, with temporal GT) | m_vIoU (%): 39.12 -> 44.12 | Improved |
| add the temporal grounding branch (with temporal GT, spatial quality) (ablation) | STVGBert · VidSTG declarative (ablation, with temporal GT) | m_vIoU (%): 45.69 -> 47.25 | Improved |
| use one-stage STVGBert instead of two-stage baselines | WSSTG + L-Net · VidSTG declarative | m_vIoU (%): 14.45 -> 23.97 | Improved |
| use STVGBert instead of STGRN | STGRN · VidSTG declarative | m_vIoU (%): 19.75 -> 23.97 | Improved |
| use STVGBert instead of STGVT | STGVT · HC-STVG | m_vIoU (%): 18.15 -> 20.42 | Improved |
Later papers that compared with this method
3 later accepted papers report STVGBert as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Large-scale Pre-training for Grounded Video Caption Generation (ICCV 2025 · better on 1 of 1 shared result)
- Embracing Consistency: A One-Stage Approach for Spatio-Temporal Video Grounding (NeurIPS 2022 · better on 6 of 6 shared results)
- TubeDETR: Spatio-Temporal Video Grounding With Transformers (CVPR 2022 · better on 3 of 3 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
- Context-Guided Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.92)
- Knowing Your Target: Target-Aware Transformer Makes Better Spatio-Temporal Video Grounding (ICLR 2025 · similarity 0.92)
- Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding (NeurIPS 2025 · similarity 0.90)
- STPro: Spatial and Temporal Progressive Learning for Weakly Supervised Spatio-Temporal Grounding (CVPR 2025 · similarity 0.90)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.88)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 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.
