Research map / Video / Video-language understanding
Video Activity Localisation with Uncertainties in Temporal Boundary
Jiabo Huang, Hailin Jin, Shaogang Gong, Yang Liu
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 |
|---|---|---|---|
| use EMB (DET) with elastic-boundary training and guided attention instead of VSLNet | VSLNet · TACoS | IoU@0.5: 27.59 -> 37.82 | Improved |
| use EMB (DET) instead of VSLNet | VSLNet · Charades-STA | IoU@0.5: 50.46 -> 58.33 | Improved |
| use EMB (DET) instead of VSLNet | VSLNet · ActivityNet-Captions | IoU@0.5: 41.10 -> 44.81 | Improved |
| use EMB (DET) instead of VSLNet | VSLNet · ActivityNet-Captions | IoU@0.7: 25.58 -> 26.07 | No clear effect |
| predict elastic boundary (ELA) instead of determined boundary at inference | EMB · TACoS | IoU@0.5: 37.82 -> 52.49 | Improved |
| evaluate EMB elastic boundary versus VSLNet with global candidate shifting | VSLNet multi-candidate · TACoS | IoU@0.5: 30.09 -> 52.49 | Improved |
| add global candidate shifting (10% of moment length) to VSLNet evaluation | VSLNet · Charades-STA | IoU@0.5: 50.46 -> 57.98 | Improved |
| add elastic moment bounding objective (component A) to the baseline (ablation) | VSLNet · Charades-STA and ActivityNet-Captions | localisation accuracy | Improved |
| add multi-grained interaction network with guided attention (component B) (ablation) | VSLNet · Charades-STA and ActivityNet-Captions | localisation accuracy | Improved |
| extend boundary heuristically (Extend) instead of single-frame endpoints (ablation) | VSLNet · Charades-STA and ActivityNet-Captions | localisation accuracy | Hurt |
| smooth boundary with gaussian kernel instead of single-frame endpoints (ablation) | VSLNet · Charades-STA and ActivityNet-Captions | localisation accuracy | Mixed |
| use Sigmoid evolving threshold instead of constant threshold for elastic boundary (ablation) | EMB · Charades-STA | localisation accuracy | Improved |
Closest papers by meaning
- Cross-Sentence Temporal and Semantic Relations in Video Activity Localisation (ICCV 2021 · similarity 0.90)
- Boundary Denoising for Video Activity Localization (ICLR 2024 · similarity 0.89)
- METAL: Minimum Effort Temporal Activity Localization in Untrimmed Videos (CVPR 2020 · similarity 0.89)
- Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization (AAAI 2025 · similarity 0.88)
- HR-Pro: Point-Supervised Temporal Action Localization via Hierarchical Reliability Propagation (AAAI 2024 · similarity 0.87)
- Energy vs. Noise: Towards Robust Temporal Action Localization in Open-World (AAAI 2025 · 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.
