Research map / Video / Action recognition
ZSTAD: Zero-Shot Temporal Activity Detection
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 ZSTAD network with label embeddings and its proposed components rather than R-C3D+SE | R-C3D · THUMOS'14 unseen activities, IoU 0.5 | mAP (%): 5.11 -> 9.15 | Helped |
| Use the ZSTAD network with label embeddings and its proposed components rather than R-C3D+ConSE | R-C3D · Charades unseen activities | standard mAP (%): 5.67 -> 7.91 | Helped |
| Add super-class clustering loss (ablation) | ZSTAD without TPN* · THUMOS'14 unseen activities, IoU 0.5 | mAP (%): 7.37 -> 8.88 | Helped |
| Add super-class clustering loss (ablation) | ZSTAD with TPN* · Charades unseen activities | standard mAP (%): 7.03 -> 7.91 | Helped |
| Add the background-embedding TPN* classification module (ablation) | ZSTAD without super-class clustering loss · THUMOS'14 unseen activities, IoU 0.5 | mAP (%): 7.37 -> 8.25 | Helped |
| Add the background-embedding TPN* classification module (ablation) | ZSTAD with super-class clustering loss · THUMOS'14 unseen activities, IoU 0.4 | mAP (%): 11.23 -> 11.12 | Mixed |
| Tune the TPN* trade-off parameter λ (ablation) | ZSTAD with TPN* · THUMOS'14 unseen activities, IoU 0.5 | mAP (%) | Helped |
| Tune the super-class loss trade-off β within its useful range (ablation) | ZSTAD with super-class clustering loss · THUMOS'14 unseen activities, IoU 0.5 | mAP (%) | Mixed |
| Apply the manually designed post-processing | ZSTAD · Charades unseen activities | mAP (%): 7.91 -> 13.23 | Helped |
Closest papers by meaning
- Text-Infused Attention and Foreground-Aware Modeling for Zero-Shot Temporal Action Detection (NeurIPS 2024 · similarity 0.89)
- DASZL: Dynamic Action Signatures for Zero-shot Learning (AAAI 2021 · similarity 0.88)
- METAL: Minimum Effort Temporal Activity Localization in Untrimmed Videos (CVPR 2020 · similarity 0.87)
- Fast Temporal Activity Proposals for Efficient Detection of Human Actions in Untrimmed Videos (CVPR 2016 · similarity 0.87)
- Learning Latent Super-Events to Detect Multiple Activities in Videos (CVPR 2018 · similarity 0.87)
- Test-Time Zero-Shot Temporal Action Localization (CVPR 2024 · 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.