Meta-Dataset

Dataset Information
Modalities
Images
Introduced
2019
License
Multiple licenses
Homepage
Official Website
Contents

Overview

The Meta-Dataset benchmark is a large few-shot learning benchmark and consists of multiple datasets of different data distributions. It does not restrict few-shot tasks to have fixed ways and shots, thus representing a more realistic scenario. It consists of 10 datasets from diverse domains:

  • ILSVRC-2012 (the ImageNet dataset, consisting of natural images with 1000 categories)
  • Omniglot (hand-written characters, 1623 classes)
  • Aircraft (dataset of aircraft images, 100 classes)
  • CUB-200-2011 (dataset of Birds, 200 classes)
  • Describable Textures (different kinds of texture images with 43 categories)
  • Quick Draw (black and white sketches of 345 different categories)
  • Fungi (a large dataset of mushrooms with 1500 categories)
  • VGG Flower (dataset of flower images with 102 categories),
  • Traffic Signs (German traffic sign images with 43 classes)
  • MSCOCO (images collected from Flickr, 80 classes).

All datasets except Traffic signs and MSCOCO have a training, validation and test split (proportioned roughly into 70%, 15%, 15%). The datasets Traffic Signs and MSCOCO are reserved for testing only.

Source: Optimized Generic Feature Learning for Few-shot Classification across Domains
Image Source: Triantafillou et al

Variants: Meta-Dataset, Meta-Dataset Rank

Associated Benchmarks

This dataset is used in 1 benchmark:

Recent Benchmark Submissions

Task Model Paper Date
Few-Shot Image Classification TSP (ResNet18; applied on TA^2-Net) Task-Specific Preconditioner for Cross-Domain Few-Shot … 2024-12-20
Few-Shot Image Classification SMAT (DINO-VIT-Base-16-224) Unleashing the Power of Meta-tuning … 2024-03-13
Few-Shot Image Classification UpperCaSE-ResNet50 Contextual Squeeze-and-Excitation for Efficient Few-Shot … 2022-06-20
Few-Shot Image Classification UpperCaSE-EfficientNetB0 Contextual Squeeze-and-Excitation for Efficient Few-Shot … 2022-06-20
Few-Shot Image Classification P>M>F (P=DINO-ViT-base, M=ProtoNet) Pushing the Limits of Simple … 2022-04-15
Few-Shot Image Classification TSA (ResNet18, URL, residual adapters, 84x84 image, shuffled data, scratch, MDL) Cross-domain Few-shot Learning with Task-specific … 2021-07-01
Few-Shot Image Classification URL (ResNet18, 84x84 image, shuffled data, scratch, MDL) Universal Representation Learning from Multiple … 2021-03-25
Few-Shot Image Classification Invariance-Equivariance Exploring Complementary Strengths of Invariant … 2021-03-01
Few-Shot Image Classification URT+MQDA Shallow Bayesian Meta Learning for … 2021-01-08
Few-Shot Image Classification URT A Universal Representation Transformer Layer … 2020-06-21
Few-Shot Image Classification Transductive CNAPS Enhancing Few-Shot Image Classification with … 2020-06-17
Few-Shot Image Classification SUR Selecting Relevant Features from a … 2020-03-20
Few-Shot Image Classification SUR-pnf Selecting Relevant Features from a … 2020-03-20
Few-Shot Image Classification Simple CNAPS Improved Few-Shot Visual Classification 2019-12-07
Few-Shot Image Classification CNAPs Fast and Flexible Multi-Task Classification … 2019-06-18
Few-Shot Image Classification k-NN Meta-Dataset: A Dataset of Datasets … 2019-03-07
Few-Shot Image Classification fo-Proto-MAML Meta-Dataset: A Dataset of Datasets … 2019-03-07
Few-Shot Image Classification Finetune Meta-Dataset: A Dataset of Datasets … 2019-03-07
Few-Shot Image Classification Relation Networks Learning to Compare: Relation Network … 2017-11-16
Few-Shot Image Classification Prototypical Networks Prototypical Networks for Few-shot Learning 2017-03-15

Research Papers

Recent papers with results on this dataset:

External Links: