Research map / Video / Action recognition
Completeness Modeling and Context Separation for Weakly Supervised Temporal Action Localization
Check 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 the proposed multi-branch network and hard-negative mining | weakly supervised temporal action localization with I3D features · THUMOS’14 test | average mAP (IoU 0.1:0.5): 39.8 -> 40.9 | Helped |
| use the proposed multi-branch network and hard-negative mining | weakly supervised temporal action localization with UntrimmedNet features · THUMOS’14 test | average mAP (IoU 0.1:0.5): 33.7 -> 37.4 | Helped |
| use the proposed method | weakly supervised temporal action localization with I3D features · ActivityNet1.2 validation | average mAP (IoU 0.5:0.05:0.95): 18.0 -> 22.4 | Helped |
| use the proposed method | weakly supervised temporal action localization with I3D features · ActivityNet1.3 test | average mAP (IoU 0.5:0.05:0.95): 20.1 -> 23.1 | Helped |
| add hard-negative videos (ablation) | single-branch model with MIL loss · THUMOS’14 test | average mAP (IoU 0.1:0.5): 28.8 -> 32.7 | Helped |
| use multiple classification branches and Lsum instead of a single branch with MIL loss (ablation) | weakly supervised temporal action localization · THUMOS’14 test | average mAP (IoU 0.1:0.5): 28.8 -> 34.8 | Helped |
| add diversity loss (ablation) | multiple-branch model trained with MIL loss and hard negatives · THUMOS’14 test | average mAP (IoU 0.1:0.5): 34.7 -> 35.6 | Helped |
| add norm regularization (ablation) | multiple-branch model with diversity loss and hard negatives · THUMOS’14 test | average mAP (IoU 0.1:0.5): 35.6 -> 37.4 | Helped |
| add the temporal attention module (ablation) | multiple-branch full model with hard negatives · THUMOS’14 test | average mAP (IoU 0.1:0.5): 36.3 -> 37.4 | Helped |
| vary the hard-negative selection ratio among 15%, 20%, 25%, and 30% (ablation) | single-branch model with hard-negative mining · THUMOS’14 test | average mAP (IoU 0.1:0.5): 32.4, 33.3, 32.7, 32.8 | No clear effect |
Closest papers by meaning
- ACSNet: Action-Context Separation Network for Weakly Supervised Temporal Action Localization (AAAI 2021 · similarity 0.93)
- Two-Stream Networks for Weakly-Supervised Temporal Action Localization With Semantic-Aware Mechanisms (CVPR 2023 · similarity 0.92)
- Relational Prototypical Network for Weakly Supervised Temporal Action Localization (AAAI 2020 · similarity 0.92)
- Action Unit Memory Network for Weakly Supervised Temporal Action Localization (CVPR 2021 · similarity 0.92)
- Weakly Supervised Action Localization by Sparse Temporal Pooling Network (CVPR 2018 · similarity 0.91)
- Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Feature (AAAI 2024 · similarity 0.91)
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.