Research map / Video / Action recognition
Revisiting the Spatial and Temporal Modeling for Few-Shot 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 FFAS to the baseline (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 42.0 -> 42.8 | Helped |
| Add LTMM and STMM to the baseline (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 42.0 -> 43.6 | Helped |
| Combine FFAS with LTMM and STMM (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 42.0 -> 46.5 | Helped |
| Use architecture search over Sum, GP_low, and GP_high (ablation) | FFAS · SSV2, 5-way 1-shot | accuracy: 45.2 -> 46.5 | Helped |
| Use GP_high rather than no FFAS fusion (ablation) | FFAS · SSV2, 5-way 1-shot | accuracy: 43.6 -> 44.5 | Helped |
| Use only STMM for temporal modeling (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 42.8 -> 43.3 | Helped |
| Use only LTMM for temporal modeling (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 42.8 -> 43.7 | Helped |
| Combine STMM and LTMM with learnable parallel weighted summation (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 45.7 -> 46.5 | Helped |
| Use Ω={1} rather than Ω={1,2} for prototype matching (ablation) | SloshNet · Kinetics, 5-way 5-shot | accuracy: 86.7 -> 87.0 | Helped |
| Use Ω={1,2} rather than Ω={1} for prototype matching (ablation) | SloshNet · SSV2, 5-way 1-shot | accuracy: 46.3 -> 46.5 | Helped |
| Integrate low-level spatial features and long- and short-term temporal features | SloshNet versus TRX · Kinetics, 5-way 5-shot | accuracy: +1.1% | Helped |
| Add short-term temporal modeling to spatial-temporal modeling | SloshNet versus STRM · Kinetics, 5-way 5-shot | accuracy: +0.5% | Helped |
Closest papers by meaning
- Spatio-Temporal Relation Modeling for Few-Shot Action Recognition (CVPR 2022 · similarity 0.90)
- Searching for Better Spatio-temporal Alignment in Few-Shot Action Recognition (NeurIPS 2022 · similarity 0.89)
- StNet: Local and Global Spatial-Temporal Modeling for Action Recognition (AAAI 2019 · similarity 0.88)
- Spatio-Temporal Naive-Bayes Nearest-Neighbor (ST-NBNN) for Skeleton-Based Action Recognition (CVPR 2017 · similarity 0.86)
- Frame Order Matters: A Temporal Sequence-Aware Model for Few-Shot Action Recognition (AAAI 2025 · similarity 0.86)
- Building a Multi-modal Spatiotemporal Expert for Zero-shot Action Recognition with CLIP (AAAI 2025 · similarity 0.85)
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.