Research map / Video / Action recognition
Boosting Few-shot Action Recognition with Graph-guided Hybrid Matching
Jiazheng Xing, Mengmeng Wang, Yudi Ruan, Bofan Chen, Yaowei Guo, Boyu Mu, Guang Dai, Jingdong Wang, Yong Liu
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 LDTM module to baseline (ablation) | GgHM baseline · SSv2 5-way 1-shot | accuracy: 44.6 -> 51.8 | Improved |
| add GgPC module to baseline (ablation) | GgHM baseline · SSv2 5-way 1-shot | accuracy: 44.6 -> 49.0 | Improved |
| add HPM hybrid matching to baseline (ablation) | GgHM baseline · SSv2 5-way 1-shot | accuracy: 44.6 -> 45.2 | Improved |
| combine all three modules (ablation) | GgHM · SSv2 5-way 5-shot | accuracy: 56.0 -> 69.2 | Improved |
| add temporal channel relation modeling to PTRM (ablation) | LDTM · SSv2 5-way 1-shot | accuracy: 53.1 -> 54.5 | Improved |
| use label-supervised GNN instead of unsupervised self-attention for task features (ablation) | GgPC · SSv2 5-way 1-shot | accuracy: 53.7 -> 54.0 | Improved |
| add transduction (all queries in one graph) to GNN (ablation) | GgPC · SSv2 5-way 1-shot | accuracy: 54.0 -> 54.5 | Improved |
| build any task-oriented features (self-attention) versus none (ablation) | Prototype construction · Kinetics 5-way 1-shot | accuracy: 72.9 -> 74.1 | Improved |
| hybrid matching instead of frame-level matching (ablation) | HPM · Kinetics 5-way 1-shot | accuracy: 74.3 -> 74.9 | Improved |
| tuple-level matching alone instead of frame-level on Kinetics (ablation) | HPM · Kinetics 5-way 1-shot | accuracy: 74.3 -> 74.1 | Hurt |
| vary hybrid fusion weight alpha (ablation) | HPM · Kinetics 5-way 1-shot | accuracy: 74.3 -> 74.9 (alpha 0.4) | Mixed |
| GgHM versus HyRSM | GgHM · SSv2 5-way 1-shot | accuracy: 51.5 -> 54.5 | Improved |
Closest papers by meaning
- Matching Compound Prototypes for Few-Shot Action Recognition (IJCV 2024 · similarity 0.90)
- MPL: Match-guided Prototype Learning for Few-shot Action Recognition (CVPR 2026 · similarity 0.90)
- Hybrid Relation Guided Set Matching for Few-Shot Action Recognition (CVPR 2022 · similarity 0.90)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.88)
- Frame Order Matters: A Temporal Sequence-Aware Model for Few-Shot Action Recognition (AAAI 2025 · similarity 0.87)
- Prompt-guided Disentangled Representation for Action Recognition (NeurIPS 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.
