Research map / Video / Video-language understanding
Are Binary Annotations Sufficient? Video Moment Retrieval via Hierarchical Uncertainty-Based Active Learning
Wei Ji, Renjie Liang, Zhedong Zheng, Wenqiao Zhang, Shengyu Zhang, Juncheng Li, Mengze Li, Tat-seng Chua
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 binary expert frame annotations using HUAL to the CPL-based baseline | HUAL on Charades-STA · Charades-STA | mIoU: 43.76 -> 48.11 | Improved |
| Add binary expert frame annotations using HUAL to the CPL-based baseline | HUAL on ActivityNet Captions · ActivityNet Captions | mIoU: 36.78 -> 40.86 | Improved |
| Increase annotation rounds from 2 to 5 while selecting 50% of videos (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: 46.63 -> 48.11 | Improved |
| Increase annotation rounds from 2 to 5 while selecting 50% of videos (ablation) | HUAL on ActivityNet Captions · ActivityNet Captions | mIoU: 38.11 -> 40.86 | Improved |
| Add model probability to distance probability for frame uncertainty (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: 36.82 -> 43.5 | Improved |
| Add previous-round frame labels to distance and model probabilities (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: 43.5 -> 48.11 | Improved |
| Increase selected video share from 10% to 50% (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: 45.46 -> 48.11 | Improved |
| Increase selected video share from 50% to 100% (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: 48.11 -> 49.70 | Improved |
| Increase selected video share from 50% to 100% (ablation) | HUAL on Charades-STA · Charades-STA | R@0.7: 28.90 -> 32.69 | Improved |
| Use more annotation rounds (ablation) | HUAL on Charades-STA · Charades-STA | mIoU: about 0.9% on average | Improved |
Closest papers by meaning
- MomentSeeker: A Task-Oriented Benchmark For Long-Video Moment Retrieval (NeurIPS 2025 · similarity 0.84)
- Hierarchical Multi-Source Uncertainty Aggregation for Interactive Video Captioning (AAAI 2025 · similarity 0.84)
- Generative Video Diffusion for Unseen Novel Semantic Video Moment Retrieval (AAAI 2025 · similarity 0.82)
- VERIFIED: A Video Corpus Moment Retrieval Benchmark for Fine-Grained Video Understanding (NeurIPS 2024 · similarity 0.82)
- Learnability Matters: Active Learning for Video Captioning (NeurIPS 2024 · similarity 0.81)
- From Play to Replay: Composed Video Retrieval for Temporally Fine-Grained Videos (NeurIPS 2025 · similarity 0.81)
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.
