Research map / Video / Action recognition
Temporal Action Detection Using a Statistical Language Model
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 |
|---|---|---|---|
| add a language model to the action detector (ablation) | Temporal action detection on Thumos 14 · Thumos 14 test | mAP at overlap 0.1: 0.285 -> 0.397 | Helped |
| add a length model to the action detector (ablation) | Temporal action detection on Thumos 14 · Thumos 14 test | mAP at overlap 0.1: 0.209 -> 0.397 | Helped |
| use a trigram rather than a unigram language model (ablation) | Temporal action detection · Thumos 14 test | mAP at overlap 0.1: 0.195 -> 0.397 | Helped |
| use a trigram rather than a bigram language model (ablation) | Temporal action detection · Thumos 14 test | mAP at overlap 0.1: 0.366 -> 0.397 | Helped |
| use a non-uniform class prior in the action model (ablation) | Temporal action detection on Thumos 14 · Thumos 14 test | mAP at overlap 0.1: 0.397 -> 0.310 | Hurt |
| remove the segment-length power factor from the action model (ablation) | Temporal action detection on Thumos 14 · Thumos 14 test | mAP at overlap 0.1: 0.397 -> 0.334 | Hurt |
| use a class-dependent Poisson length model instead of the mean length model (ablation) | Temporal action detection · Thumos 14 | mAP at overlap 0.1: 0.397 -> 0.251 | Hurt |
| use a class-dependent Poisson length model instead of the mean length model (ablation) | Temporal action detection · MPII-Cooking | mAP at overlap 0.1: 0.220 -> 0.248 | Helped |
| use a class-dependent Poisson length model instead of the mean length model (ablation) | Temporal action detection · 50 Salads low level | mAP at overlap 0.1: 0.305 -> 0.379 | Helped |
| subsample frames at 20 rather than 5 frames (ablation) | Temporal action detection · 50 Salads low level | mAP at overlap 0.1: 0.391 -> 0.376 | Hurt |
Closest papers by meaning
- Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints (CVPR 2018 · similarity 0.90)
- OnlineTAS: An Online Baseline for Temporal Action Segmentation (NeurIPS 2024 · similarity 0.89)
- ASM-Loc: Action-Aware Segment Modeling for Weakly-Supervised Temporal Action Localization (CVPR 2022 · similarity 0.89)
- Unsupervised Action Segmentation via Fast Learning of Semantically Consistent Actoms (AAAI 2024 · similarity 0.88)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 2024 · similarity 0.87)
- Benchmarking the Robustness of Temporal Action Detection Models Against Temporal Corruptions (CVPR 2024 · 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.