Research map / Video / Action recognition
Unsupervised Few-Shot Action Recognition via Action-Appearance Aligned Meta-Adaptation
Jay Patravali, Gaurav Mittal, Ye Yu, Fuxin Li, Mei Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 the appearance stream to the action stream (no A3M) (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 53.56 -> 59.16 | Improved |
| add the action stream to the appearance stream (no A3M) (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 50.91 -> 59.16 | Improved |
| add the A3M module (ablation) | MetaUVFS · UCF101 5-way 1-shot | accuracy: 73.02 -> 76.38 | Improved |
| add the A3M module (ablation) | MetaUVFS · HMDB51 5-way 1-shot | accuracy: 44.89 -> 47.55 | Improved |
| add the A3M module (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 59.16 -> 62.80 | Improved |
| add the A3M module (ablation) | MetaUVFS · HMDB51 5-way 5-shot | accuracy: 64.96 -> 66.13 | Improved |
| add hard-mined episodes (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 59.88 -> 62.80 | Improved |
| add hard-mined episodes (ablation) | MetaUVFS · UCF101 5-way 1-shot | accuracy: 73.97 -> 76.38 | Improved |
| add hard-mined episodes without A3M (ablation) | MetaUVFS without A3M · Kinetics100 5-way 1-shot | accuracy: 57.52 -> 59.16 | Improved |
| use MAML instead of ProtoMAML (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 62.13 -> 62.80 | Improved |
| use MAML instead of ProtoNet (ablation) | MetaUVFS · Kinetics100 5-way 1-shot | accuracy: 31.20 -> 62.80 | Improved |
| use action + appearance streams instead of two appearance streams (ablation) | MetaUVFS without A3M · Kinetics100 5-way 1-shot | accuracy: 55.75 -> 59.16 | Improved |
Closest papers by meaning
- Few-Shot Video Classification via Representation Fusion and Promotion Learning (ICCV 2023 · similarity 0.89)
- A Comprehensive Review of Few-Shot Action Recognition (IJCV 2025 · similarity 0.89)
- Domain-Specific Priors and Meta Learning for Few-Shot First-Person Action Recognition (TPAMI 2021 · similarity 0.87)
- Few-Shot Action Recognition with Hierarchical Matching and Contrastive Learning (ECCV 2022 · similarity 0.87)
- CDFSL-V: Cross-Domain Few-Shot Learning for Videos (ICCV 2023 · similarity 0.87)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.86)
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.
