Research map / Video / Action recognition
Learning Temporal Co-Attention Models for Unsupervised Video 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 |
|---|---|---|---|
| Iteratively use class-agnostic attention to weight video features during clustering, then retrain localization using the resulting pseudo-labels | ACL on THUMOS14 · THUMOS14 test | mAP@0.5: 13.0 -> 20.8 -> 25.0 | Helped |
| Add the Local-Global feature aggregation block (ablation) | ACL model with classification loss on THUMOS14 · THUMOS14 test | Avg(0.3:0.7): 21.39 -> 22.16 | Helped |
| Add action-background separation loss (ablation) | ACL model with Local-Global feature aggregation and classification loss · THUMOS14 test | Avg(0.3:0.7): 22.16 -> 23.48 | Helped |
| Add cluster-based triplet loss (ablation) | ACL model with Local-Global feature aggregation and classification loss · THUMOS14 test | Avg(0.3:0.7): 22.16 -> 23.05 | Helped |
| Use both action-background separation loss and cluster-based triplet loss (ablation) | ACL model with Local-Global feature aggregation and classification loss · THUMOS14 test | Avg(0.3:0.7): 22.16 -> 24.68 | Helped |
| Use unsupervised action co-localization rather than weak video-level labels | Proposed method on THUMOS14 · THUMOS14 test | mAP@0.5: 30.1 -> 25.0 | Hurt |
| Use unsupervised action co-localization rather than weak video-level labels | Proposed method on ActivityNet-1.2 · ActivityNet-1.2 validation | Average mAP: 24.6 -> 21.1 | Hurt |
Closest papers by meaning
- ACSNet: Action-Context Separation Network for Weakly Supervised Temporal Action Localization (AAAI 2021 · similarity 0.93)
- Weakly Supervised Temporal Action Localization Through Learning Explicit Subspaces for Action and Context (AAAI 2021 · similarity 0.92)
- Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Feature (AAAI 2024 · similarity 0.91)
- Bridge the Gap: From Weak to Full Supervision for Temporal Action Localization with PseudoFormer (CVPR 2025 · similarity 0.90)
- Realigning Confidence with Temporal Saliency Information for Point-Level Weakly-Supervised Temporal Action Localization (CVPR 2024 · similarity 0.90)
- Temporal Action Localization with Cross Layer Task Decoupling and Refinement (AAAI 2025 · similarity 0.90)
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.