Research map / Video / Action recognition
Global2Local: Efficient Structure Search for Video Action Segmentation
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 |
|---|---|---|---|
| Search receptive-field combinations with Global2Local | MS-TCN · BreakFast | F@0.1: 52.6 -> 74.9 | Helped |
| Search receptive-field combinations with Global2Local | MS-TCN · 50Salads | F@0.1: 76.3 -> 80.3 | Helped |
| Apply searched receptive-field structure | MS-TCN++ · BreakFast | F@0.1: 64.1 -> 72.4 | Helped |
| Apply searched receptive-field structure | BCN · BreakFast | F@0.1: 68.7 -> 72.5 | Helped |
| Use searched structure with SSTDA | SSTDA · BreakFast | F@0.1: 75.0 -> 76.3 | Helped |
| Add EGI local search after global search | MS-TCN · GTEA | F@0.1: 89.1 -> 89.9 | Helped |
| Add EGI local search | MS-TCN · BreakFast | F@0.1: 73.8 -> 74.9 | Helped |
| Initialize EGI local search from a globally searched structure rather than a random structure (ablation) | MS-TCN · BreakFast | F@0.1: 73.6 -> 74.9 | Helped |
| Initialize EGI local search from a globally searched structure rather than the baseline structure (ablation) | MS-TCN · BreakFast | F@0.1: 74.1 -> 74.9 | Helped |
| Vary the number of sampled dilation rates from S=3 to S=2 (ablation) | EGI local search on MS-TCN · BreakFast | F@0.1: 74.9 -> 74.8 | No clear effect |
| Vary the number of sampled dilation rates from S=3 to S=4 (ablation) | EGI local search on MS-TCN · BreakFast | F@0.1: 74.9 -> 74.9 | No clear effect |
| Replace absolute-weight normalization with sigmoid (ablation) | EGI local search on MS-TCN · BreakFast | F@0.1: 74.9 -> 72.7 | Hurt |
Closest papers by meaning
- Unsupervised Action Segmentation via Fast Learning of Semantically Consistent Actoms (AAAI 2024 · similarity 0.85)
- OnlineTAS: An Online Baseline for Temporal Action Segmentation (NeurIPS 2024 · similarity 0.85)
- Your Image is My Video: Reshaping the Receptive Field via Image-To-Video Differentiable AutoAugmentation and Fusion (CVPR 2024 · similarity 0.85)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.84)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 2024 · similarity 0.84)
- ActFusion: a Unified Diffusion Model for Action Segmentation and Anticipation (NeurIPS 2024 · 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.