Research map / Video / Action recognition
Inductive and Transductive Few-Shot Video Classification via Appearance and Temporal Alignments
Khoi D. Nguyen, Quoc-Huy Tran, Khoi Nguyen, Binh-Son Hua, Rang Nguyen
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 |
|---|---|---|---|
| use appearance+temporal score instead of optimal transport (ablation) | Ours (inductive) · Kinetics 5-way 1-shot | accuracy (%): 61.71 -> 74.26 | Improved |
| use log-sum-exp score instead of max operator (ablation) | Ours (inductive) · SSv2 5-way 1-shot | accuracy (%): 40.97 -> 43.82 | Improved |
| use log-sum-exp score instead of max operator (ablation) | Ours (inductive) · Kinetics 5-way 5-shot | accuracy (%): 87.67 -> 87.40 | No clear effect |
| add InfoNCE loss to supervised loss (ablation) | Ours (inductive) · SSv2 5-way 5-shot | accuracy (%): 57.26 -> 61.07 | Improved |
| add InfoNCE loss to supervised loss (ablation) | Ours (inductive) · Kinetics 5-way 1-shot | accuracy (%): 74.25 -> 74.26 | No clear effect |
| add InfoNCE loss to supervised loss (ablation) | Ours (transductive) · SSv2 5-way 5-shot | accuracy (%): 67.56 -> 69.42 | Improved |
| add transductive prototype refinement (ablation) | Ours · Kinetics 5-way 1-shot | accuracy (%): 74.26 -> 83.08 | Improved |
| add transductive prototype refinement (ablation) | Ours · SSv2 5-way 5-shot | accuracy (%): 61.07 -> 69.42 | Improved |
| add temporal term (beta=0.4) to appearance-only score (ablation) | Ours (inductive) · SSv2 5-way 1-shot | accuracy (%): 48.13 -> 48.55 | Improved |
| add temporal term (beta=0.3) to appearance-only score (ablation) | Ours (transductive) · SSv2 5-way 5-shot | accuracy (%): 73.97 -> 75.47 | Improved |
| add temporal term (small beta) to appearance-only score (ablation) | Ours (inductive) · Kinetics 5-way 1-shot | accuracy (%): 75.37 -> 75.38 | No clear effect |
| use temporal-only score (beta=1.0) (ablation) | Ours (inductive) · Kinetics 5-way 1-shot | accuracy (%): 75.37 -> 21.09 | Hurt |
Closest papers by meaning
- Few-Shot Video Classification via Temporal Alignment (CVPR 2020 · similarity 0.91)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.88)
- Generalized Few-Shot Video Classification With Video Retrieval and Feature Generation (TPAMI 2021 · similarity 0.87)
- Temporal Transductive Inference for Few-Shot Video Object Segmentation (IJCV 2025 · similarity 0.87)
- Searching for Better Spatio-temporal Alignment in Few-Shot Action Recognition (NeurIPS 2022 · similarity 0.87)
- Frame Order Matters: A Temporal Sequence-Aware Model for Few-Shot Action Recognition (AAAI 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.
