ML Research Wiki / Benchmarks / Image Classification / OmniBenchmark

OmniBenchmark

Image Classification Benchmark

Performance Over Time

📊 Showing 22 results | 📏 Metric: Average Top-1 Accuracy

Top Performing Models

Rank	Model	Paper	Average Top-1 Accuracy	Date	Code
1	NOAH-ViTB/16 📚	Neural Prompt Search	47.60	2022-06-09	📦 ZhangYuanhan-AI/NOAH
2	SwinTransformer	Swin Transformer: Hierarchical Vision Transformer using Shifted Windows	46.40	2021-03-25	📦 huggingface/transformers 📦 rwightman/pytorch-image-models 📦 open-mmlab/mmdetection
3	Bamboo-R50 📚	Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy	45.40	2022-03-15	📦 zhangyuanhan-ai/bamboo 📦 davidzhangyuanhan/bamboo
4	Adapter-ViTB/16 📚	Parameter-Efficient Transfer Learning for NLP	44.50	2019-02-02	📦 Adapter-Hub/adapter-transformers 📦 CyberZHG/keras-bert 📦 kpe/bert-for-tf2
5	CLIP-RN50 📚	Learning Transferable Visual Models From Natural Language Supervision	42.10	2021-02-26	📦 openai/CLIP 📦 mlfoundations/open_clip 📦 towhee-io/towhee
6	IG-1B 📚	Billion-scale semi-supervised learning for image classification	40.40	2019-05-02	📦 facebookresearch/semi-supervised-ImageNet1K-models 📦 leaderj1001/Billion-scale-semi-supervised-learning 📦 salesforce/ensemble-of-averages 📦 tiskw/patchcore-ad
7	BiT-M 📚	Big Transfer (BiT): General Visual Representation Learning	40.40	2019-12-24	📦 google-research/big_transfer 📦 sayakpaul/FunMatch-Distillation 📦 bethgelab/InDomainGeneralizationBenchmark
8	DINO	Emerging Properties in Self-Supervised Vision Transformers	38.90	2021-04-29	📦 facebookresearch/dino 📦 lightly-ai/lightly 📦 facebookresearch/vissl
9	SwAV	Unsupervised Learning of Visual Features by Contrasting Cluster Assignments	38.30	2020-06-17	📦 open-mmlab/mmdetection 📦 lightly-ai/lightly 📦 facebookresearch/vissl
10	ResNet-101	Deep Residual Learning for Image Recognition	37.40	2015-12-10	📦 tensorflow/models 📦 tensorflow/models 📦 tensorflow/models

All Papers (22)

Neural Prompt Search

2022

NOAH-ViTB/16

ZhangYuanhan-AI/NOAH

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

2021

SwinTransformer

huggingface/transformers rwightman/pytorch-image-models

Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy

2022

Bamboo-R50

zhangyuanhan-ai/bamboo davidzhangyuanhan/bamboo

Parameter-Efficient Transfer Learning for NLP

2019

Adapter-ViTB/16

Adapter-Hub/adapter-transformers CyberZHG/keras-bert

Learning Transferable Visual Models From Natural Language Supervision

2021

CLIP-RN50

openai/CLIP mlfoundations/open_clip

Billion-scale semi-supervised learning for image classification

2019

IG-1B

facebookresearch/semi-supervised-ImageNet1K-models leaderj1001/Billion-scale-semi-supervised-learning

Big Transfer (BiT): General Visual Representation Learning

2019

BiT-M

google-research/big_transfer sayakpaul/FunMatch-Distillation

Emerging Properties in Self-Supervised Vision Transformers

2021

DINO

facebookresearch/dino lightly-ai/lightly

Unsupervised Learning of Visual Features by Contrasting Cluster Assignments

2020

SwAV

open-mmlab/mmdetection lightly-ai/lightly

Deep Residual Learning for Image Recognition

2015

ResNet-101

tensorflow/models tensorflow/models

MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1 Accuracy on ImageNet without Tricks

2020

MEAL-V2

szq0214/MEAL-V2

MoPro: Webly Supervised Learning with Momentum Prototypes

2020

MoPro-V2

salesforce/MoPro yuleiqin/capro

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

2019

EfficientNetB4

ultralytics/yolov5 rwightman/pytorch-image-models

Momentum Contrast for Unsupervised Visual Representation Learning

2019

MoCoV2

open-mmlab/mmdetection KevinMusgrave/pytorch-metric-learning

Deep Residual Learning for Image Recognition

2015

ResNet-50

tensorflow/models tensorflow/models

Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning

2016

InceptionV4

tensorflow/models tensorflow/models

MLP-Mixer: An all-MLP Architecture for Vision

2021

MLP-Mixer

labmlai/annotated_deep_learning_paper_implementations rwightman/pytorch-image-models

Manifold Mixup: Better Representations by Interpolating Hidden States

2018

Manifold

Westlake-AI/openmixup vikasverma1077/manifold_mixup

CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features

2019

CutMix

rwightman/pytorch-image-models pytorch/vision

Re-labeling ImageNet: from Single to Multi-Labels, from Global to Localized Labels

2021

ReLabel

naver-ai/relabel_imagenet aiiu-lab/naturalistic-adversarial-patch

Masked Autoencoders Are Scalable Vision Learners

2021

MAE

facebookresearch/mae lightly-ai/lightly

BEiT: BERT Pre-Training of Image Transformers

2021

BeiT

huggingface/transformers rwightman/pytorch-image-models

OmniBenchmark

Performance Over Time

Edit Benchmark Results

Edit Result

Top Performing Models

All Papers (22)

Neural Prompt Search

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy

Parameter-Efficient Transfer Learning for NLP

Learning Transferable Visual Models From Natural Language Supervision

Billion-scale semi-supervised learning for image classification

Big Transfer (BiT): General Visual Representation Learning

Emerging Properties in Self-Supervised Vision Transformers

Unsupervised Learning of Visual Features by Contrasting Cluster Assignments

Deep Residual Learning for Image Recognition

MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1 Accuracy on ImageNet without Tricks

MoPro: Webly Supervised Learning with Momentum Prototypes

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

Momentum Contrast for Unsupervised Visual Representation Learning

Deep Residual Learning for Image Recognition

Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning

MLP-Mixer: An all-MLP Architecture for Vision

Manifold Mixup: Better Representations by Interpolating Hidden States

CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features

Re-labeling ImageNet: from Single to Multi-Labels, from Global to Localized Labels

Masked Autoencoders Are Scalable Vision Learners

BEiT: BERT Pre-Training of Image Transformers

Model	Paper	Average Top-1 Accuracy	Date
NOAH-ViTB/16	Neural Prompt Search	47.60	2022-06-09
SwinTransformer	Swin Transformer: Hierarchical Vision Transformer…	46.40	2021-03-25
Bamboo-R50	Bamboo: Building Mega-Scale Vision Dataset Contin…	45.40	2022-03-15
Adapter-ViTB/16	Parameter-Efficient Transfer Learning for NLP	44.50	2019-02-02
CLIP-RN50	Learning Transferable Visual Models From Natural …	42.10	2021-02-26
IG-1B	Billion-scale semi-supervised learning for image …	40.40	2019-05-02
BiT-M	Big Transfer (BiT): General Visual Representation…	40.40	2019-12-24
DINO	Emerging Properties in Self-Supervised Vision Tra…	38.90	2021-04-29
SwAV	Unsupervised Learning of Visual Features by Contr…	38.30	2020-06-17
ResNet-101	Deep Residual Learning for Image Recognition	37.40	2015-12-10
MEAL-V2	MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1…	36.60	2020-09-17
MoPro-V2	MoPro: Webly Supervised Learning with Momentum Pr…	36.10	2020-09-17
EfficientNetB4	EfficientNet: Rethinking Model Scaling for Convol…	35.80	2019-05-28
MoCoV2	Momentum Contrast for Unsupervised Visual Represe…	34.80	2019-11-13
ResNet-50	Deep Residual Learning for Image Recognition	34.30	2015-12-10
InceptionV4	Inception-v4, Inception-ResNet and the Impact of …	32.30	2016-02-23
MLP-Mixer	MLP-Mixer: An all-MLP Architecture for Vision	32.20	2021-05-04
Manifold	Manifold Mixup: Better Representations by Interpo…	31.60	2018-06-13
CutMix	CutMix: Regularization Strategy to Train Strong C…	31.10	2019-05-13
ReLabel	Re-labeling ImageNet: from Single to Multi-Labels…	30.80	2021-01-13
MAE	Masked Autoencoders Are Scalable Vision Learners	30.60	2021-11-11
BeiT	BEiT: BERT Pre-Training of Image Transformers	30.10	2021-06-15