Research map / Video / Video-language understanding
Natural Language Video Localization: A Revisit in Span-based Question Answering Framework
Hao Zhang, Aixin Sun, Wei Jing, Liangli Zhen, Joey Tianyi Zhou, Rick Siow Mong Goh
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 query-guided highlighting (VSLNet over VSLBase) (ablation) | VSLBase · Charades-STA (no fine-tune) | Rank@1 IoU=0.7: 24.14 -> 30.19 | Improved |
| add query-guided highlighting (ablation) | VSLBase · Charades-STA (no fine-tune) | mIoU: 42.11 -> 45.15 | Improved |
| add query-guided highlighting (ablation) | VSLBase · Charades-STA (fine-tuned) | mIoU: 47.15 -> 50.02 | Improved |
| fine-tune the 3D ConvNet on Charades | VSLNet · Charades-STA | Rank@1 IoU=0.7: 30.19 -> 35.22 | Improved |
| replace CAT with context-query attention (keep CMF encoder) (ablation) | VSLBase · Charades-STA | Rank@1 IoU=0.7: 27.07 -> 30.16 | Improved |
| replace BiLSTM with CMF encoder (keep CQA) (ablation) | VSLBase · Charades-STA | Rank@1 IoU=0.7: 28.55 -> 30.16 | Improved |
| add multi-scale split-and-concat (VSLNet-L-Pm) over VSLNet | VSLNet · TACoS_tan (n=600) | mIoU: 31.92 -> 36.61 | Improved |
| add multi-scale split-and-concat (VSLNet-L-Pm) over VSLNet | VSLNet · TACoS_org (n=600) | mIoU: 23.96 -> 26.40 | Improved |
| use multi-scale rather than single-scale split (ablation) | VSLNet-L · TACoS_org (n=600) | mIoU: 25.12 -> 26.40 | Improved |
| use multi-scale rather than single-scale split (ablation) | VSLNet-L · TACoS_tan (n=600) | mIoU: 33.71 -> 36.61 | Improved |
| add split-and-concat to videos shorter than 2 minutes | VSLNet · TACoS_tan (n=600) 0-2 min | mIoU: 46.73 -> 44.32 | Hurt |
| add split-and-concat on videos longer than 12 min | VSLNet · TACoS_org (n=600) >12 min | mIoU: 8.58 -> 11.90 | Improved |
Closest papers by meaning
- Boundary Proposal Network for Two-stage Natural Language Video Localization (AAAI 2021 · similarity 0.92)
- Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization (AAAI 2025 · similarity 0.89)
- Commonsense for Zero-Shot Natural Language Video Localization (AAAI 2024 · similarity 0.88)
- Localizing Events in Videos with Multimodal Queries (CVPR 2025 · similarity 0.88)
- Transferable Video Moment Localization by Moment-Guided Query Prompting (AAAI 2024 · similarity 0.87)
- Unsupervised Domain Adaptative Temporal Sentence Localization with Mutual Information Maximization (AAAI 2024 · similarity 0.87)
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.
