Research map / Video / Action recognition
Temporal Action Localization by Structured Maximal Sums
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 start-middle-end temporal evolution modeling (ablation) | Ours (full) · THUMOS '14 test, σ=0.5 | mAP: 16.2 -> 17.8 | Helped |
| add the structured classification loss (ablation) | Ours (full) · THUMOS '14 test, σ=0.5 | mAP: 15.7 -> 17.8 | Helped |
| add both structured classification loss and start-middle-end temporal evolution (ablation) | framewise-classification baseline · THUMOS '14 test, σ=0.5 | mAP: 0.2 -> 17.8 | Helped |
| add action duration priors (ablation) | Ours (full) · THUMOS '14 test, σ=0.5 | mAP: 17.5 -> 17.8 | Helped |
| jointly fine-tune the spatial and motion streams instead of using late fusion (ablation) | two-stream model · THUMOS '14 test, σ=0.5 | mAP: 14.5 -> 17.8 | Helped |
| combine jointly trained spatial and motion streams rather than use the spatial stream alone (ablation) | two-stream model · THUMOS '14 test, σ=0.5 | mAP: 16.0 -> 17.8 | Helped |
| combine jointly trained spatial and motion streams rather than use the motion stream alone (ablation) | two-stream model · THUMOS '14 test, σ=0.5 | mAP: 16.9 -> 17.8 | Helped |
| use structured maximal sums with structured training instead of framewise classification training (ablation) | THUMOS '14 localization · THUMOS '14 test, σ=0.1 | mAP: 18.5 -> 51.0 | Helped |
| use structured maximal sums with structured training instead of framewise classification training (ablation) | THUMOS '14 localization · THUMOS '14 test, σ=0.5 | mAP: 0.2 -> 17.8 | Helped |
| use the full method rather than Shou et al. | THUMOS '14 localization · THUMOS '14 test, σ=0.1 | mAP: 47.7 -> 51.0 | Helped |
| use the full method rather than Shou et al. | THUMOS '14 localization · THUMOS '14 test, σ=0.2 | mAP: 43.5 -> 45.2 | Helped |
| use the full method rather than Shou et al. | THUMOS '14 localization · THUMOS '14 test, σ=0.3 | mAP: 36.3 -> 36.5 | Helped |
Closest papers by meaning
- Temporal Action Localization in Untrimmed Videos via Multi-Stage CNNs (CVPR 2016 · similarity 0.93)
- CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos (CVPR 2017 · similarity 0.92)
- Rethinking the Faster R-CNN Architecture for Temporal Action Localization (CVPR 2018 · similarity 0.91)
- One-Shot Action Localization by Learning Sequence Matching Network (CVPR 2018 · similarity 0.91)
- Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Feature (AAAI 2024 · similarity 0.90)
- Realigning Confidence with Temporal Saliency Information for Point-Level Weakly-Supervised Temporal Action Localization (CVPR 2024 · similarity 0.89)
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.