Research map / Video / Video-language understanding
OmniSTVG: Toward Spatio-Temporal Omni-Object Video Grounding
Jiali Yao, Xin Gu, Xinran Deng, Mengrui Dai, Bing Fan, Zhipeng Zhang, Yan Huang, Heng Fan, Libo Zhang
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 spatial and temporal attention blocks (SAB, TAB) to queries (ablation) | OmniTube spatial decoder · BOSTVG test (Full) | m_vIoU: 8.25 -> 9.00 (m_tIoU 34.33 -> 34.13) | Improved |
| add spatial query generation (SQG) with either attention block (ablation) | OmniTube spatial decoder · BOSTVG test (Full) | m_tIoU, m_vIoU: m_tIoU 34.33 -> 34.98/35.42; m_vIoU 8.25 -> 8.89/9.15 | Improved |
| use all of SQG, SAB and TAB together (ablation) | OmniTube spatial decoder · BOSTVG test (Full) | m_tIoU: 34.33 -> 35.83 | Improved |
| add temporal attention block to temporal decoder (ablation) | OmniTube temporal decoder · BOSTVG test (Full) | m_tIoU: 26.06 -> 35.00 | Improved |
| add temporal query generation alone (ablation) | OmniTube temporal decoder · BOSTVG test (Full) | m_tIoU: 26.06 -> 26.00 | No clear effect |
| predict word position index instead of box class (ablation) | OmniTube · BOSTVG test (Full) | m_vIoU: 8.82 -> 9.47 | Improved |
| add VidSwin motion features (ablation) | OmniTube · BOSTVG test (Full) | m_vIoU: 8.88 -> 9.47 | Improved |
| set M (top-M features for query generation) to 5 versus 2 or 10 (ablation) | OmniTube · BOSTVG test (Full) | m_vIoU: 9.14 (M=2), 9.47 (M=5), 9.34 (M=10) | Improved |
| OmniTube versus adapted TubeDETR, STCAT, CG-STVG | adapted single-object STVG methods · BOSTVG test (Full) | m_vIoU: 7.66 (CG-STVG) -> 9.47 | Improved |
| OmniTube versus STCAT on High-target subset | STCAT · BOSTVG test (High) | vIoU@0.3: 1.30 -> 1.30 | No clear effect |
| full OmniTube versus baseline without query generation and attention blocks (ablation) | Baseline · BOSTVG test (Full) | m_vIoU: 5.38 -> 9.47 | 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 7 of 7 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 0.91)
- Knowing Your Target: Target-Aware Transformer Makes Better Spatio-Temporal Video Grounding (ICLR 2025 · similarity 0.90)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.89)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 2025 · similarity 0.89)
- Unleashing the Potential of Multimodal LLMs for Zero-Shot Spatio-Temporal Video Grounding (NeurIPS 2025 · similarity 0.89)
- STPro: Spatial and Temporal Progressive Learning for Weakly Supervised Spatio-Temporal Grounding (CVPR 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.
