Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Fine-grained and cross-domain few-shot baselines
Uses baselines, feature reconstruction and prototypical refinement for fine-grained or cross-domain few-shot tasks.
59 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Zero-/few-shot and meta-learning, keyword cluster “fsl · shot classification · support”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
17 papers in 2023–24 and 11 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 8.3% to 6.2%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · One-Shot Learning of Scene Locations via Feature Trajectory Transfer (CVPR 2016 · 51 citations)
- 2017 · Multi-Attention Network for One Shot Learning (CVPR 2017 · 93 citations)
- 2018 · MetaGAN: An Adversarial Approach to Few-Shot Learning (NeurIPS 2018 · 277 citations)
- 2019 · Coloring With Limited Data: Few-Shot Colorization via Memory Augmented Networks (CVPR 2019 · 147 citations)
- 2020 · Diversity Transfer Network for Few-Shot Learning (AAAI 2020 · 74 citations)
- 2021 · Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need? (CVPR 2021 · 188 citations)
- 2022 · Task Discrepancy Maximization for Fine-Grained Few-Shot Classification (CVPR 2022 · 93 citations)
- 2023 · Revisiting Prototypical Network for Cross Domain Few-Shot Learning (CVPR 2023 · 98 citations)
- 2024 · Cross-Layer and Cross-Sample Feature Optimization Network for Few-Shot Fine-Grained Image Classification (AAAI 2024 · 55 citations)
- 2025 · MVREC: A General Few-shot Defect Classification Model Using Multi-View Region-Context (AAAI 2025 · 9 citations)
- 2026 · F2SST: Frequency-to-Spatial Semantic Transfer for Few-Shot Image Classification (AAAI 2026 · 0 citations)
Earliest papers here
- One-Shot Learning of Scene Locations via Feature Trajectory Transfer (CVPR 2016 · 51 citations)
- Multi-Attention Network for One Shot Learning (CVPR 2017 · 93 citations)
- MetaGAN: An Adversarial Approach to Few-Shot Learning (NeurIPS 2018 · 277 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
CUB 5-way 1-shot · accuracy (higher is better)
7 papers, 2019–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 84.8 | Meta-Reinforced Synthetic Data for One-Shot Fine-Grained Visual Recognition (NeurIPS 2019 · 13 citations) |
| 2023 | 85.44 | Bi-directional Feature Reconstruction Network for Fine-Grained Few-Shot Image Classification (AAAI 2023 · 71 citations) |
CUB 5-way 5-shot · accuracy (higher is better)
7 papers, 2019–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 93.09 | Meta-Reinforced Synthetic Data for One-Shot Fine-Grained Visual Recognition (NeurIPS 2019 · 13 citations) |
| 2023 | 94.73 | Bi-directional Feature Reconstruction Network for Fine-Grained Few-Shot Image Classification (AAAI 2023 · 71 citations) |
miniImageNet 5-way 1-shot · accuracy (higher is better)
5 papers, 2018–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 52.71 | MetaGAN: An Adversarial Approach to Few-Shot Learning (NeurIPS 2018 · 277 citations) |
| 2019 | 58.35 | A Dual Attention Network with Semantic Embedding for Few-Shot Learning (AAAI 2019 · 60 citations) |
| 2020 | 63.45 | Diversity Transfer Network for Few-Shot Learning (AAAI 2020 · 74 citations) |
| 2023 | 68.72 | RankDNN: Learning to Rank for Few-Shot Learning (AAAI 2023 · 28 citations) |
miniImageNet 5-way 5-shot · accuracy (higher is better)
5 papers, 2018–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 68.63 | MetaGAN: An Adversarial Approach to Few-Shot Learning (NeurIPS 2018 · 277 citations) |
| 2019 | 71.07 | A Dual Attention Network with Semantic Embedding for Few-Shot Learning (AAAI 2019 · 60 citations) |
| 2020 | 77.91 | Diversity Transfer Network for Few-Shot Learning (AAAI 2020 · 74 citations) |
| 2023 | 85.66 | RankDNN: Learning to Rank for Few-Shot Learning (AAAI 2023 · 28 citations) |
What has been tried
51 papers here report 587 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 6%, hurt 19%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add CLFR to the baseline | CUB (raw images), 1-shot · accuracy (%) | 54.87 -> 69.81 | Helped | Cross-Layer and Cross-Sample Feature Optimization Network for Few-Shot Fine-Grained Image Classification (AAAI 2024 · 55 citations) |
| Add MPG to the baseline | ISIC 1-shot · mIoU | 44.55 -> 36.67 | Hurt | APSeg: Auto-Prompt Network for Cross-Domain Few-Shot Semantic Segmentation (CVPR 2024 · 34 citations) |
| use task adaptation before dense comparison | CD-FSS average 1-shot · mIoU | +5.86 | Helped | Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentation (CVPR 2024 · 25 citations) |
| Add BFP to baseline SSP for source training and fine-tune without query labels | Deepglobe · mIoU (%) | 41.3 -> 44.1 | Helped | Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining (CVPR 2024 · 24 citations) |
| Add brightness FroFA | ILSVRC-2012 1-, 5-, and 10-shot · top-1 accuracy | +4.8; +1.1; +0.6 | Helped | Frozen Feature Augmentation for Few-Shot Image Classification (CVPR 2024 · 22 citations) |
| Replace prototype-aware cross-attention with vanilla cross-attention | COCO-20i · mIoU (%) | 31.0 -> 29.4 | Hurt | Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · 16 citations) |
| Use MVREC features instead of the unaugmented representation | MVTec-FS, 1-shot · accuracy | +11.3 | Helped | MVREC: A General Few-shot Defect Classification Model Using Multi-View Region-Context (AAAI 2025 · 9 citations) |
| Add PFRM to the backbone | CUB-200-2011, Conv-4 1-shot · accuracy | 68.87 -> 77.54 | Helped | Few-Shot Fine-Grained Image Classification with Progressively Feature Refinement and Continuous Relationship Modeling (AAAI 2025 · 5 citations) |
| Replace the zero-shot CLIP predictor prior mean with a constant-zero or learnable-vector mean | ImageNet low-shot classification · accuracy | Hurt | Bayesian Exploration of Pre-trained Models for Low-shot Image Classification (CVPR 2024 · 4 citations) | |
| Add auxiliary loss to base-class pre-training | PASCAL-5^i 1-shot · H-Mean | 48.58 -> 50.29 | Helped | Enhancing Generalized Few-Shot Semantic Segmentation via Effective Knowledge Transfer (AAAI 2025 · 3 citations) |
| add dynamic bias rectification instead of static bias rectification | Multi-source dataset 1-shot · accuracy | 35.47±0.25* -> 35.93±1.82* | Helped | Manhattan Self-Attention Diffusion Residual Networks with Dynamic Bias Rectification for BCI-based Few-Shot Learning (AAAI 2025 · 1 citation) |
| Add base-rate adjustment to the evidential model | Cifar100 1-shot · ECE | 0.499±0.004 -> 0.077±0.004 | Helped | Be Confident in What You Know: Bayesian Parameter Efficient Fine-Tuning of Vision Foundation Models (NeurIPS 2024 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 7 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CUB-200-2011 · 2 of 7 papers
Baselines
- ProtoNet · 2 of 7 papers
- FEAT · 2 of 7 papers
- DeepEMD · 2 of 7 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 28 papers; from OpenAlex and the papers' affiliations).
- University of Technology Sydney · 3 papers
- Zhejiang University · 3 papers
- Fudan University · 2 papers
- Shandong University · 2 papers
- Beijing Institute of Technology · 2 papers
- Shenzhen University · 2 papers
Use it
Download BibTeX of 39 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Embedding adaptation for few-shot classification (71 papers)
- Fine-tuning strategies for few-shot (67 papers)
- Alignment-based few-shot metric learning (38 papers)
Elsewhere in the topic
- Generative feature synthesis for zero-shot (117 papers)
- Zero-shot robustness and transfer (101 papers)
- Robust fine-tuning of zero-shot models (92 papers)
- Transductive and semantic-dictionary zero-shot (90 papers)
- Meta-augmentation and meta-objectives (60 papers)
- Meta-learning analysis and benchmarks (50 papers)
- Normalization and unsupervised meta-learning (39 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).