Research map / Video / Action recognition
Multi-Label Activity Recognition Using Activity-Specific Features and Activity Correlations
Yanyi Zhang, Xinyu Li, Ivan Marsic
Check 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 activity-specific features and activity correlation | CSN-152 · Charades val | mAP: 46.4 -> 50.3 | Improved |
| add activity-specific features and activity correlation | SlowFast-101 · Charades val | mAP: 44.7 -> 48.1 | Improved |
| add activity-specific features and activity correlation | Slowfast-101 · AVA val | mAP @ 0.5 IoU: 29.1 -> 30.2 | Improved |
| add activity-specific features and activity correlation | CSN-152 · Volleyball personal | accuracy: 85.0 -> 86.6 | Improved |
| set group size to 64 rather than 32 (ablation) | Our's · Charades val | mAP: 48.4 -> 47.2 | Hurt |
| increase group size from 1 to 32 (ablation) | Our's · Charades val | mAP: 47.9 -> 48.4 | No clear effect |
| increase observations from 16 to 64 (ablation) | Our's · Charades val | mAP: 46.1 -> 48.4 | Improved |
| increase observations from 64 to 128 (ablation) | Our's · Charades val | mAP: 48.4 -> 48.4 | No clear effect |
| add activity-specific features (ablation) | Our's · Charades val | mAP: 48.4 -> 49.2 | Improved |
| add activity correlation (ablation) | Our's · Charades val | mAP: 48.4 -> 49.2 | Improved |
| combine sampling rates 2 and 4 instead of using rate 4 only (ablation) | Our's · Charades val | mAP: 49.2 -> 49.9 | Improved |
| combine sampling rates 2, 4, and 8 instead of rates 2 and 4 (ablation) | Our's · Charades val | mAP: 49.9 -> 50.3 | Improved |
Closest papers by meaning
- Joint Prediction of Activity Labels and Starting Times in Untrimmed Videos (ICCV 2017 · similarity 0.89)
- LabelForest: Non-Parametric Semi-Supervised Learning for Activity Recognition (AAAI 2019 · similarity 0.87)
- Weakly-Guided Self-Supervised Pretraining for Temporal Activity Detection (AAAI 2023 · similarity 0.86)
- Detector-Free Weakly Supervised Group Activity Recognition (CVPR 2022 · similarity 0.85)
- Discriminative Hierarchical Rank Pooling for Activity Recognition (CVPR 2016 · similarity 0.85)
- Domain-Informed Label Fusion Surpasses LLMs in Free-Living Activity Classification (Student Abstract) (AAAI 2025 · similarity 0.84)
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.
