Research map / Video / Action recognition
Every Moment Counts: Dense Detailed Labeling of Actions in Complex Videos
Serena Y. Yeung, Olga Russakovsky, Ning Jin, Mykhaylo Andriluka, Greg Mori, Li Fei-Fei
Read the paperCheck 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 input connections with averaging (ablation) | LSTM · MultiTHUMOS | mAP: 28.1 -> 28.7 | Improved |
| add soft attention over input window (ablation) | LSTM + i · MultiTHUMOS | mAP: 28.7 -> 29.1 | Improved |
| add output connections (ablation) | LSTM + i + a · MultiTHUMOS | mAP: 29.1 -> 29.7 | Improved |
| add input connections with averaging (ablation) | LSTM · THUMOS | mAP: 39.3 -> 39.5 | Improved |
| add soft attention over input window (ablation) | LSTM + i · THUMOS | mAP: 39.5 -> 39.7 | Improved |
| add output connections (ablation) | LSTM + i + a · THUMOS | mAP: 39.7 -> 41.3 | Improved |
| use MultiLSTM instead of plain LSTM (ablation) | LSTM · MultiTHUMOS | mAP: 28.1 -> 29.7 | Improved |
| add LSTM temporal modeling on top of single-frame CNN (ablation) | Single-frame CNN · MultiTHUMOS | mAP: 25.4 -> 28.1 | Improved |
| replace hand-crafted IDT with fine-tuned CNN features (ablation) | IDT · MultiTHUMOS | mAP: 13.3 -> 25.4 | Improved |
| increase attention units from 50 to beyond 75 (ablation) | MultiLSTM · MultiTHUMOS | mAP | Hurt |
| add soft attention over output predictions (ablation) | MultiLSTM · MultiTHUMOS | mAP | No clear effect |
Closest papers by meaning
- Multi-Moments in Time: Learning and Interpreting Models for Multi-Action Video Understanding (TPAMI 2021 · similarity 0.91)
- An Effective-Efficient Approach for Dense Multi-Label Action Detection (IJCV 2026 · similarity 0.89)
- Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (CVPR 2019 · similarity 0.89)
- No Frame Left Behind: Full Video Action Recognition (CVPR 2021 · similarity 0.89)
- Weakly-Supervised Temporal Action Detection for Fine-Grained Videos with Hierarchical Atomic Actions (ECCV 2022 · similarity 0.88)
- Deep Sequential Context Networks for Action Prediction (CVPR 2017 · similarity 0.88)
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.
