Research map / Video / Video segmentation and tracking
Learning Latent Super-Events to Detect Multiple Activities in Videos
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 |
|---|---|---|---|
| Add soft-attention super-event representation | I3D · MultiTHUMOS | mAP: 29.7 -> 36.4 | Helped |
| Add soft-attention super-event representation | I3D · Charades v1 localization | mAP: 17.22 -> 19.41 | Helped |
| Add soft-attention super-event representation | I3D Two-Stream · AVA test set | mAP: 7.5 -> 9.8 | Helped |
| Replace global max pooling with learned super-events (ablation) | I3D · MultiTHUMOS | mAP: 30.0 -> 36.4 | Helped |
| Replace global mean pooling with learned super-events (ablation) | I3D · MultiTHUMOS | mAP: 30.8 -> 36.4 | Helped |
| Replace temporal pyramid pooling with learned super-events (ablation) | I3D · MultiTHUMOS | mAP: 31.2 -> 36.4 | Helped |
| Add a single per-class super-event representation (ablation) | I3D baseline · MultiTHUMOS | mAP: 29.7 -> 31.2 | Helped |
| Use shared temporal structure filters with per-class soft attention instead of one per-class super-event (ablation) | I3D · MultiTHUMOS | mAP: 31.2 -> 36.4 | Helped |
| Use relative super-events instead of global super-events (ablation) | I3D with soft attention · MultiTHUMOS | mAP: 36.4 -> 36.2 | Hurt |
| Add LSTM temporal modeling | I3D · MultiTHUMOS | mAP: 29.7 -> 29.9 | Helped |
| Add LSTM temporal modeling | I3D · Charades v1 localization | mAP: 17.22 -> 18.1 | Helped |
Closest papers by meaning
- Title Learning Latent Subevents in Activity Videos Using Temporal Attention Filters (AAAI 2017 · similarity 0.92)
- Unsupervised Learning of Action Classes With Continuous Temporal Embedding (CVPR 2019 · similarity 0.88)
- Fast Temporal Activity Proposals for Efficient Detection of Human Actions in Untrimmed Videos (CVPR 2016 · similarity 0.88)
- End-To-End Learning of Action Detection From Frame Glimpses in Videos (CVPR 2016 · similarity 0.88)
- ZSTAD: Zero-Shot Temporal Activity Detection (CVPR 2020 · similarity 0.87)
- Weakly-Guided Self-Supervised Pretraining for Temporal Activity Detection (AAAI 2023 · similarity 0.87)
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.