Research map / Video / Action recognition
SMART Frame Selection for Action Recognition
Read the paperCheck 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 GloVe language features to visual features (ablation) | SMART global selector · Something-something-v2 | accuracy: 58.3 -> 59.2 | Helped |
| Combine the single-frame selector with the global selector (ablation) | SMART · Something-something-v2 | accuracy: 59.7 -> 60.6 | Helped |
| Use pairs of frames as global-selector input (ablation) | SMART · Something-something-v2 | accuracy: 60.6 -> 60.8 | Helped |
| Select frames with SMART instead of uniform sampling | ResNet-152 action recognition · UCF101 | accuracy: 69.1 -> 75.3 | Helped |
| Select frames with SMART instead of using all frames | ResNet-152 action recognition · UCF101 | accuracy: 74.6 -> 75.3 | Helped |
| Use SMART-selected frames rather than all frames | ResNet-152 action recognition · Kinetics-Temporal | accuracy: 64.1 -> 65.4 | Helped |
| Use SMART-selected frames rather than all frames | ResNet-152 action recognition · Kinetics-Static | accuracy: 62.4 -> 62.6 | Helped |
| Use SMART as a frame-selection preprocessor instead of AAS | TSN · UCF101 | accuracy: 94.6 -> 95.8 | Helped |
| Use SMART as a frame-selection preprocessor instead of AAS | TSN · HMDB51 | accuracy: 71.2 -> 74.6 | Helped |
| Use SMART-selected frames at inference | ISTPAN · UCF101 | accuracy: 95.5 -> 96.4 | Helped |
| Use SMART-selected frames at inference | I3D · HMDB51 | accuracy: 80.0 -> 81.1 | Helped |
Closest papers by meaning
- AdaFrame: Adaptive Frame Selection for Fast Video Recognition (CVPR 2019 · similarity 0.90)
- Search-Map-Search: A Frame Selection Paradigm for Action Recognition (CVPR 2023 · similarity 0.89)
- No Frame Left Behind: Full Video Action Recognition (CVPR 2021 · similarity 0.88)
- Are all Frames Equal? Active Sparse Labeling for Video Action Detection (NeurIPS 2022 · similarity 0.87)
- LookWhen? Fast Video Recognition by Learning When, Where, and What to Compute (NeurIPS 2026 · similarity 0.87)
- ActionAtlas: A VideoQA Benchmark for Domain-specialized Action Recognition (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.