Research map / Video / Action recognition
Fast Learning of Temporal Action Proposal via Dense Boundary Generator
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 DBG instead of BMN | temporal proposal generation · ActivityNet-1.3 val | AUC: 67.10 -> 68.23 | Helped |
| use DBG instead of BMN | temporal proposal generation · ActivityNet-1.3 test | AUC: 67.19 -> 68.57 | Helped |
| add temporal boundary classification maps rather than predicting boundary probability sequences (ablation) | DBG · ActivityNet-1.3 val | AUC | Helped |
| add action-aware completeness features and auxiliary actionness supervision instead of using dual-stream features for completeness regression (ablation) | DBG · ActivityNet-1.3 val | AUC | Helped |
| use the dual-stream BaseNet rather than concatenated spatial and temporal features with a BSN-like BaseNet (ablation) | DBG · ActivityNet-1.3 val | AUC | Helped |
| sample 8/16/8 points from left/center/right proposal regions rather than 4/8/4 (ablation) | PFG · ActivityNet-1.3 val | AUC: 67.91 -> 68.23 | Helped |
| add left and right context sampling to center-region sampling (ablation) | PFG · ActivityNet-1.3 val | AUC: 66.94 -> 68.23 | Helped |
| add center proposal-region features to left and right local-region features (ablation) | PFG · ActivityNet-1.3 val | AUC: 67.83 -> 68.23 | Helped |
| increase PFG sampling from 8/16/8 to 10/20/10 points (ablation) | PFG · ActivityNet-1.3 val | AUC: 68.23 -> 68.11 | Hurt |
| use DBG proposals instead of BMN proposals | SCNN-cls action detection · THUMOS14 test at tIoU 0.7 | mAP: 17.0 -> 18.4 | Helped |
| use DBG proposals instead of BMN proposals | UNet action detection · THUMOS14 test at tIoU 0.7 | mAP: 20.5 -> 21.7 | Helped |
Closest papers by meaning
- BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation (AAAI 2021 · similarity 0.90)
- Temporal Context Aggregation Network for Temporal Action Proposal Refinement (CVPR 2021 · similarity 0.89)
- Efficient Temporal Action Segmentation via Boundary-aware Query Voting (NeurIPS 2024 · similarity 0.87)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.86)
- Condensing Action Segmentation Datasets via Generative Network Inversion (CVPR 2025 · similarity 0.85)
- HyrCap: Hybrid Rank-Calibration of Action Proposals for Temporal Event Understanding (NeurIPS 2026 · similarity 0.85)
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.