Research map / Video / Video-language understanding
Zero-Shot Natural Language Video Localization
Jinwoo Nam, Daechul Ahn, Dongyeop Kang, Seong Jong Ha, Jonghyun Choi
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 VerbBERT verb inference instead of RoBERTa (ablation) | PSVL pseudo-query · Charades-STA | R@0.5: 15.49 -> 31.29 | Improved |
| add VerbBERT verbs instead of random verbs (ablation) | PSVL pseudo-query · Charades-STA | R@0.5: 13.18 -> 31.29 | Improved |
| add video-aware words attention (VWA) (ablation) | PSVL NLVL model · Charades-STA | R@0.5: 21.81 -> 31.29 | Improved |
| add words-aware video attention (WVA) (ablation) | PSVL NLVL model · Charades-STA | R@0.5: 27.91 -> 31.29 | Improved |
| add multi-modal cross attention (MCA) (ablation) | PSVL NLVL model · Charades-STA | R@0.5: 25.8 -> 31.29 | Improved |
| use temporal event proposals instead of random temporal regions (ablation) | PSVL · Charades-STA | R@0.5: 19.06 -> 31.29 | Improved |
| use pseudo-queries instead of random queries with TEP (ablation) | PSVL · Charades-STA | R@0.5: 12.17 -> 31.29 | Improved |
| use contextualized similarity-matrix features instead of frame features for event proposal (ablation) | PSVL TEP · Charades-STA | R@0.5: 28.98 -> 31.29 | Improved |
| use uniform sampling instead of diversity scoring for composite events (ablation) | PSVL TEP · Charades-STA | R@0.5: 22.9 -> 31.29 | Improved |
| use uniform sampling instead of compactness scoring for composite events (ablation) | PSVL TEP · Charades-STA | R@0.5: 27.82 -> 31.29 | Improved |
| train with simplified nouns-and-verbs query instead of original sentences | LGI on Charades-STA · Charades-STA | R@0.5: 60.05 -> 60.22 | No clear effect |
| higher overlap of detected nouns with original nouns (1.15% -> 36.54%) | PSVL · Charades-STA | R@0.5: 23.82 -> 31.88 | Improved |
Closest papers by meaning
- Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization (AAAI 2025 · similarity 0.93)
- Commonsense for Zero-Shot Natural Language Video Localization (AAAI 2024 · similarity 0.92)
- Omnipotent Distillation with LLMs for Weakly-Supervised Natural Language Video Localization: When Divergence Meets Consistency (AAAI 2024 · similarity 0.89)
- Transferable Video Moment Localization by Moment-Guided Query Prompting (AAAI 2024 · similarity 0.89)
- Dense Video Object Captioning from Disjoint Supervision (ICLR 2025 · similarity 0.89)
- 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.
