Few-Shot Image Classification

Edit

94 papers with code • 65 benchmarks • 17 datasets

Few-shot image classification is the task of doing image classification with only a few examples for each category (typically < 6 examples).

( Image credit: Learning Embedding Adaptation for Few-Shot Learning )

Benchmarks

Add a Result

Dataset	Best Model	Paper Title	Compare
Mini-Imagenet 5-way (1-shot)	Illumination Augmentation	Sill-Net: Feature Augmentation with Separated Illumination Representation	See all
Mini-Imagenet 5-way (5-shot)	Transductive CNAPS + FETI	Enhancing Few-Shot Image Classification with Unlabelled Examples	See all
Tiered ImageNet 5-way (1-shot)	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
Tiered ImageNet 5-way (5-shot)	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
CIFAR-FS 5-way (5-shot)	Illumination Augmentation	Sill-Net: Feature Augmentation with Separated Illumination Representation	See all
CIFAR-FS 5-way (1-shot)	LST+MAP	Transfer learning based few-shot classification using optimal transport mapping from preprocessed latent space of backbone neural network	See all
CUB 200 5-way 1-shot	Illumination Augmentation	Sill-Net: Feature Augmentation with Separated Illumination Representation	See all
OMNIGLOT - 1-Shot, 20-way	GCR	Few-Shot Learning with Global Class Representations	See all
CUB 200 5-way 5-shot	Illumination Augmentation	Sill-Net: Feature Augmentation with Separated Illumination Representation	See all
OMNIGLOT - 5-Shot, 20-way	MC2+	Meta-Curvature	See all
OMNIGLOT - 1-Shot, 5-way	MC2+	Meta-Curvature	See all
OMNIGLOT - 5-Shot, 5-way	DCN6-E	Decoder Choice Network for Meta-Learning	See all
Mini-Imagenet 10-way (1-shot)	Transductive CNAPS + FETI	Enhancing Few-Shot Image Classification with Unlabelled Examples	See all
Mini-Imagenet 10-way (5-shot)	Transductive CNAPS + FETI	Enhancing Few-Shot Image Classification with Unlabelled Examples	See all
Meta-Dataset	URT	A Universal Representation Transformer Layer for Few-Shot Image Classification	See all
Mini-ImageNet - 1-Shot Learning	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
Tiered ImageNet 10-way (1-shot)	Transductive CNAPS + FETI	Enhancing Few-Shot Image Classification with Unlabelled Examples	See all
Tiered ImageNet 10-way (5-shot)	Transductive CNAPS + FETI	Enhancing Few-Shot Image Classification with Unlabelled Examples	See all
Meta-Dataset Rank	URT	A Universal Representation Transformer Layer for Few-Shot Image Classification	See all
FC100 5-way (1-shot)	R2-D2+Task Aug	Task Augmentation by Rotating for Meta-Learning	See all
FC100 5-way (5-shot)	R2-D2+Task Aug	Task Augmentation by Rotating for Meta-Learning	See all
Mini-ImageNet-CUB 5-way (1-shot)	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
Stanford Dogs 5-way (5-shot)	MML(KL)	Multi-level Metric Learning for Few-shot Image Recognition	See all
Stanford Cars 5-way (1-shot)	MATANet	Multi-scale Adaptive Task Attention Network for Few-Shot Learning	See all
Stanford Cars 5-way (5-shot)	MATANet	Multi-scale Adaptive Task Attention Network for Few-Shot Learning	See all
Mini-Imagenet 20-way (1-shot)	TIM-GD	Transductive Information Maximization For Few-Shot Learning	See all
Mini-Imagenet 20-way (5-shot)	TIM-GD	Transductive Information Maximization For Few-Shot Learning	See all
CUB-200-2011 - 0-Shot	Word CNN-RNN (DS-SJE Embedding)	Learning Deep Representations of Fine-grained Visual Descriptions	See all
Mini-Imagenet 5-way (10-shot)	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
Mini-ImageNet-CUB 5-way (5-shot)	PT+MAP	Leveraging the Feature Distribution in Transfer-based Few-Shot Learning	See all
ImageNet - 1-shot	ViT-MoE-15B (Every-2)	Scaling Vision with Sparse Mixture of Experts	See all
ImageNet - 5-shot	ViT-MoE-15B (Every-2)	Scaling Vision with Sparse Mixture of Experts	See all
CUB 200 50-way (0-shot)	Prototypical Networks	Prototypical Networks for Few-shot Learning	See all
ImageNet - 10-shot	ViT-MoE-15B (Every-2)	Scaling Vision with Sparse Mixture of Experts	See all
CUB-200 - 0-Shot Learning	TAFE-Net	TAFE-Net: Task-Aware Feature Embeddings for Low Shot Learning	See all
Stanford Dogs 5-way (1-shot)	MML(KL)	Multi-level Metric Learning for Few-shot Image Recognition	See all
SUN - 0-Shot	Synthesised Classifier	Synthesized Classifiers for Zero-Shot Learning	See all
ImageNet - 0-Shot	Synthesised Classifier	Synthesized Classifiers for Zero-Shot Learning	See all
CIFAR100 5-way (1-shot)	UL-Hopfield (ULH)	Unsupervised Learning using Pretrained CNN and Associative Memory Bank	See all
Caltech-256 5-way (1-shot)	UL-Hopfield (ULH)	Unsupervised Learning using Pretrained CNN and Associative Memory Bank	See all
Mini-ImageNet to CUB - 5 shot learning	TIM-GD	Transductive Information Maximization For Few-Shot Learning	See all
ImageNet (1-shot)	Dynamic FSL	Dynamic Few-Shot Visual Learning without Forgetting	See all
Fewshot-CIFAR100 - 1-Shot Learning	pseudo-shots	Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks	See all
Fewshot-CIFAR100 - 5-Shot Learning	pseudo-shots	Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks	See all
CIFAR-FS - 1-Shot Learning	pseudo-shots	Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks	See all
Flowers-102 - 0-Shot	Word CNN-RNN (DS-SJE Embedding)	Learning Deep Representations of Fine-grained Visual Descriptions	See all
AWA - 0-Shot	Synthesised Classifier	Synthesized Classifiers for Zero-Shot Learning	See all
OMNIGLOT-EMNIST 5-way (1-shot)	DKT + BNCosSim	Bayesian Meta-Learning for the Few-Shot Setting via Deep Kernels	See all
OMNIGLOT-EMNIST 5-way (5-shot)	DKT + BNCosSim	Bayesian Meta-Learning for the Few-Shot Setting via Deep Kernels	See all
FC100 5-way (10-shot)	MTL	Meta-Transfer Learning for Few-Shot Learning	See all
OMNIGLOT - 1-Shot, 423 way	APL	Adaptive Posterior Learning: few-shot learning with a surprise-based memory module	See all
OMNIGLOT - 5-Shot, 423 way	APL	Adaptive Posterior Learning: few-shot learning with a surprise-based memory module	See all
OMNIGLOT - 1-Shot, 1000 way	APL	Adaptive Posterior Learning: few-shot learning with a surprise-based memory module	See all
OMNIGLOT - 5-Shot, 1000 way	APL	Adaptive Posterior Learning: few-shot learning with a surprise-based memory module	See all
AWA1 - 0-Shot	TAFE-Net	TAFE-Net: Task-Aware Feature Embeddings for Low Shot Learning	See all
AWA2 - 0-Shot	TAFE-Net	TAFE-Net: Task-Aware Feature Embeddings for Low Shot Learning	See all
aPY - 0-Shot	TAFE-Net	TAFE-Net: Task-Aware Feature Embeddings for Low Shot Learning	See all
mini-ImageNet - 100-Way	GCR	Few-Shot Learning with Global Class Representations	See all
CIFAR-FS - 5-Shot Learning	pseudo-shots	Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks	See all
miniImagenet → CUB (5-way 1-shot)	LaplacianShot	Laplacian Regularized Few-Shot Learning	See all
miniImagenet → CUB (5-way 5-shot)	LaplacianShot	Laplacian Regularized Few-Shot Learning	See all
iNaturalist (227-way multi-shot)	LaplacianShot	Laplacian Regularized Few-Shot Learning	See all
CUB-200-2011 5-way (1-shot)	MATANet	Multi-scale Adaptive Task Attention Network for Few-Shot Learning	See all
CUB-200-2011 5-way (5-shot)	MATANet	Multi-scale Adaptive Task Attention Network for Few-Shot Learning	See all
Oxford 102 Flower	RS-FSL	Rich Semantics Improve Few-shot Learning	See all

Show all 65 benchmarks

Collapse benchmarks

Datasets

Subtasks

Generalized Few-Shot Classification

Greatest papers with code

Learning to Remember Rare Events

tensorflow/models • • 9 Mar 2017

We present a large-scale life-long memory module for use in deep learning.

Ranked #8 on Few-Shot Image Classification on OMNIGLOT - 5-Shot, 5-way

Few-Shot Image Classification Machine Translation +1

70,696

Paper
Code

Meta-Learning without Memorization

google-research/google-research • • ICLR 2020

If this is not done, the meta-learner can ignore the task training data and learn a single model that performs all of the meta-training tasks zero-shot, but does not adapt effectively to new image classes.

Ranked #18 on Few-Shot Image Classification on OMNIGLOT - 5-Shot, 20-way

Few-Shot Image Classification Meta-Learning

18,536

Paper
Code

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

cbfinn/maml • • ICML 2017

We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and reinforcement learning.

Ranked #2 on Few-Shot Image Classification on OMNIGLOT - 5-Shot, 5-way

Few-Shot Image Classification Few-shot Regression +2

1,958

Paper
Code

Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML

learnables/learn2learn • • ICLR 2020

We conclude with a discussion of the rapid learning vs feature reuse question for meta-learning algorithms more broadly.

Few-Shot Image Classification

1,376

Paper
Code

On First-Order Meta-Learning Algorithms

learnables/learn2learn • • 8 Mar 2018

This paper considers meta-learning problems, where there is a distribution of tasks, and we would like to obtain an agent that performs well (i. e., learns quickly) when presented with a previously unseen task sampled from this distribution.

Ranked #3 on Image Classification on Tiered ImageNet 5-way (5-shot)

Few-Shot Image Classification

1,376

Paper
Code

Prototypical Networks for Few-shot Learning

learnables/learn2learn • • NeurIPS 2017

We propose prototypical networks for the problem of few-shot classification, where a classifier must generalize to new classes not seen in the training set, given only a small number of examples of each new class.

Ranked #1 on Few-Shot Image Classification on CUB 200 50-way (0-shot)

Few-Shot Image Classification General Classification +2

1,376

Paper
Code

Learning to Compare: Relation Network for Few-Shot Learning

floodsung/LearningToCompare_FSL • • CVPR 2018

Once trained, a RN is able to classify images of new classes by computing relation scores between query images and the few examples of each new class without further updating the network.

Ranked #2 on Image Classification on Tiered ImageNet 5-way (5-shot)

Few-Shot Image Classification Zero-Shot Learning

882

Paper
Code

A Closer Look at Few-shot Classification

wyharveychen/CloserLookFewShot • • ICLR 2019

Few-shot classification aims to learn a classifier to recognize unseen classes during training with limited labeled examples.

Ranked #4 on Few-Shot Image Classification on Mini-ImageNet-CUB 5-way (5-shot)

Domain Generalization Few-Shot Image Classification +1

843

Paper
Code

Matching Networks for One Shot Learning

oscarknagg/few-shot • • NeurIPS 2016

Our algorithm improves one-shot accuracy on ImageNet from 87. 6% to 93. 2% and from 88. 0% to 93. 8% on Omniglot compared to competing approaches.

Ranked #3 on Few-Shot Image Classification on Mini-ImageNet-CUB 5-way (1-shot)

Few-Shot Image Classification Language Modelling +2

790

Paper
Code

Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes

google-research/meta-dataset • • NeurIPS 2019

We introduce a conditional neural process based approach to the multi-task classification setting for this purpose, and establish connections to the meta-learning and few-shot learning literature.

Ranked #6 on Few-Shot Image Classification on Meta-Dataset Rank

Active Learning Continual Learning +3

540

Paper
Code

Few-Shot Image Classification Edit

Benchmarks Add a Result

Datasets

Subtasks

Greatest papers with code

Few-Shot Image Classification

Edit

Benchmarks

Add a Result