Research map / Video / Video-language understanding
D3G: Exploring Gaussian Prior for Temporal Sentence Grounding with Glance Annotation
Hanjun Li, Xiujun Shu, Sunan He, Ruizhi Qiao, Wei Wen, Taian Guo, Bei Gan, Xing Sun
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 |
|---|---|---|---|
| sample a group of top-k positive moments (SA-GCL) instead of a single top-1 positive (ablation) | D3G · Charades-STA | R@1 IoU=0.5: 5.08 -> 40.51 | Improved |
| add Dynamic Gaussian prior Adjustment (DGA) (ablation) | D3G · Charades-STA | R@1 IoU=0.7: 16.10 -> 19.60 | Improved |
| add DGA (ablation) | D3G · ActivityNet Captions | R@5 IoU=0.7: 43.44 -> 52.47 | Improved |
| add DGA (ablation) | D3G · TACoS | R@1 IoU=0.5: 11.95 -> 12.67 | Improved |
| fuse Gaussian weight and semantic consistency priors for positive sampling (ablation) | SA-GCL · Charades-STA | R@1 IoU=0.5: 38.09 -> 40.51 | Improved |
| sample positives by semantic consistency alone (ablation) | SA-GCL · Charades-STA | R@1 IoU=0.5: 40.51 -> 25.67 | Hurt |
| train with extreme glance annotation near moment boundaries (ablation) | D3G vs ViGA · Charades-STA | R@1 IoU=0.5: D3G 41.64 -> 40.19; ViGA 36.56 -> 33.66 | Mixed |
| increase top-k positive moments beyond an optimal value (ablation) | SA-GCL · Charades-STA | R@1 | Hurt |
| D3G versus ViGA under glance supervision | D3G · Charades-STA | R@5 IoU=0.5: 48.90 -> 79.25 | Improved |
| D3G versus PS-VTG and PFU using I3D features on Charades-STA | D3G · Charades-STA | R@1 IoU=0.5: 41.64 vs PFU 54.66 | Hurt |
Closest papers by meaning
- Empower Words: DualGround for Structured Phrase and Sentence-Level Temporal Grounding (NeurIPS 2025 · similarity 0.90)
- Temporal Sentence Grounding with Relevance Feedback in Videos (NeurIPS 2024 · similarity 0.89)
- Temporal Sentence Grounding in Videos: A Survey and Future Directions (TPAMI 2023 · similarity 0.89)
- Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network (AAAI 2025 · similarity 0.89)
- Learning Multi-Scale Video-Text Correspondence for Weakly Supervised Temporal Article Gronding (AAAI 2024 · similarity 0.87)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.86)
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.
