CIFAR-100

Introduced by Krizhevsky et al. in Learning multiple layers of features from tiny images

The CIFAR-100 dataset (Canadian Institute for Advanced Research, 100 classes) is a subset of the Tiny Images dataset and consists of 60000 32x32 color images. The 100 classes in the CIFAR-100 are grouped into 20 superclasses. There are 600 images per class. Each image comes with a "fine" label (the class to which it belongs) and a "coarse" label (the superclass to which it belongs). There are 500 training images and 100 testing images per class.

The criteria for deciding whether an image belongs to a class were as follows:

The class name should be high on the list of likely answers to the question “What is in this picture?”
The image should be photo-realistic. Labelers were instructed to reject line drawings.
The image should contain only one prominent instance of the object to which the class refers.
The object may be partially occluded or seen from an unusual viewpoint as long as its identity is still clear to the labeler.

Source: https://www.cs.toronto.edu/~kriz/cifar.html

Homepage

Benchmarks

Add a new result Link an existing benchmark

Task	Dataset Variant	Best Model
Image Classification	CIFAR-100	EffNet-L2
Long-tail Learning	CIFAR-100-LT (ρ=100)	LPT
Long-tail Learning	CIFAR-100-LT (ρ=10)	PEL
Semi-Supervised Image Classification	cifar-100, 10000 Labels	CCSSL
Knowledge Distillation	CIFAR-100	shufflenet-v2
Image Clustering	CIFAR-100	TEMI CLIP ViT-L
Semi-Supervised Image Classification	CIFAR-100, 400 Labels	SemiReward
Incremental Learning	CIFAR-100 - 50 classes + 5 steps of 10 classes	TCIL
Anomaly Detection	One-class CIFAR-100	Transformaly
Incremental Learning	CIFAR-100 - 50 classes + 10 steps of 5 classes	TCIL
Anomaly Detection	Unlabeled CIFAR-10 vs CIFAR-100	PsudoLabels ViT
Semi-Supervised Image Classification	CIFAR-100, 2500 Labels	CCSSL
Neural Architecture Search	CIFAR-100	DNA-c
Few-Shot Class-Incremental Learning	CIFAR-100	SV-T
Continual Learning	Cifar100 (20 tasks)	Model Zoo-Continual
Class Incremental Learning	cifar100	S&B
Image Generation	CIFAR-100	LeCAM
Personalized Federated Learning	CIFAR-100	pFedGP-IP-data
Conditional Image Generation	CIFAR-100	ADC-GAN
Sparse Learning and binarization	CIFAR-100	SBNN-50%
Incremental Learning	CIFAR-100 - 50 classes + 25 steps of 2 classes	D3Former : Debiased Dual Distilled Transformer for Incremental Learning
Network Pruning	CIFAR-100	Dense
Provable Adversarial Defense	CIFAR-100	SLL X-Large
Out-of-Distribution Detection	CIFAR-100	Wide ResNet 40x2
Data Free Quantization	CIFAR-100	ResNet-20 CIFAR-100
Open-World Semi-Supervised Learning	CIFAR-100	TRSSL
Small Data Image Classification	CIFAR-100, 1000 Labels	ChimeraMix+AutoAugment
Adversarial Defense	CIFAR-100	wideresnet-34-20
Adversarial Robustness	CIFAR-100	Mixed Classifier
Semi-Supervised Image Classification	CIFAR-100, 5000Labels	LiDAM
Few-Shot Image Classification	CIFAR100 5-way (1-shot)	UL-Hopfield
Incremental Learning	CIFAR-100 - 50 classes + 50 steps of 1 class	PODNet
Stochastic Optimization	CIFAR-100	Resnet18
Classification with Binary Neural Network	CIFAR-100	BATS
Bayesian Inference	cifar100	F-SWA
Learning with coarse labels	cifar100	MaskCon
Class Incremental Learning	CIFAR-100 - 50 classes + 10 steps of 5 classes	PPCA-SWSL
Class Incremental Learning	CIFAR-100 - 50 classes + 5 steps of 10 classes	PPCA-SWSL
Self-Supervised Learning	cifar100	ResNet50
Image Classification	cifar100	shreynet
Classification	CIFAR-100	ResNet8×4
Variational Inference	cifar100	F-Deep-ens
Binarization	CIFAR-100	WRN 28-10
Self-Supervised Learning	CIFAR-100	CorInfomax
Non-exemplar-based Class Incremental Learning	cifar100	NAPA-VQ
Classification with Binary Weight Network	CIFAR-100	WRN 28-10
Semi-Supervised Image Classification	CIFAR-100, 1000 Labels	EnAET
Novel Class Discovery	cifar100	AutoNovel
Learning with noisy labels	CIFAR-100	InstanceGM
Classifier calibration	CIFAR-100	R-Mix
Image Classification with Label Noise	CIFAR-100	PropMix
Out-of-Distribution Detection	cifar100	Wide Resnet 40x2