Research map / Video / Action recognition
Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization
Linjiang Huang, Liang Wang, Hongsheng Li
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 hybrid attention to CW branch alone (ablation) | CW branch · THUMOS14 | mAP AVG (%): 31.2 -> 34.6 | Improved |
| add hybrid attention to CA branch alone (ablation) | CA branch · THUMOS14 | mAP AVG (%): 30.5 -> 32.5 | Improved |
| add hybrid attention to MIL branch alone (ablation) | MIL branch · THUMOS14 | mAP AVG (%): 29.6 -> 32.5 | Improved |
| add CW branch to CA branch (ablation) | CA branch · THUMOS14 | mAP AVG (%): 30.5 -> 35.9 | Improved |
| add CW branch to MIL branch (ablation) | MIL branch · THUMOS14 | mAP AVG (%): 29.6 -> 34.4 | Improved |
| add hybrid attention to CW+CA+MIL (ablation) | FAC-Net w/o BG · THUMOS14 | mAP AVG (%): 37.5 -> 38.8 | Improved |
| add background class to full model with hybrid attention (ablation) | FAC-Net · THUMOS14 | mAP AVG (%): 38.8 -> 42.2 | Improved |
| add background class to CW+CA+MIL without hybrid attention (ablation) | FAC-Net · THUMOS14 | mAP AVG (%): 37.5 -> 40.8 | Improved |
| use temperature 10.0 in hybrid attention (1.0, 10.0) (ablation) | FAC-Net · THUMOS14 | mAP AVG (%): 41.9 -> 33.8 | Hurt |
| use three attentions (1.0, 2.0, 5.0) instead of one (ablation) | FAC-Net · THUMOS14 | mAP AVG (%): 40.8 -> 42.2 | Improved |
| use four attentions instead of three (ablation) | FAC-Net · THUMOS14 | mAP AVG (%): 42.2 -> 42.0 | No clear effect |
| plug CW branch into STPN + Embedding | STPN · THUMOS14 | mAP AVG (%): 28.9 -> 30.6 | Improved |
Closest papers by meaning
- Cross-Task Relation-Aware Consistency for Weakly Supervised Temporal Action Detection (TPAMI 2025 · similarity 0.91)
- A Hybrid Attention Mechanism for Weakly-Supervised Temporal Action Localization (AAAI 2021 · similarity 0.91)
- D2-Net: Weakly-Supervised Action Localization via Discriminative Embeddings and Denoised Activations (ICCV 2021 · similarity 0.91)
- Weakly-Supervised Action Localization With Background Modeling (ICCV 2019 · similarity 0.90)
- Two-Branch Relational Prototypical Network for Weakly Supervised Temporal Action Localization (TPAMI 2021 · similarity 0.90)
- Similar Modality Enhancement and Action Consistency Learning for Weakly Supervised Temporal Action Localization (AAAI 2025 · similarity 0.89)
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.
