ML Research Wiki / Benchmarks / Data Augmentation / ImageNet

ImageNet

Data Augmentation Benchmark

Performance Over Time

📊 Showing 17 results | 📏 Metric: Accuracy (%)

Top Performing Models

Rank	Model	Paper	Accuracy (%)	Date	Code
1	DeiT-B (+MixPro)	MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer	82.90	2023-04-24	📦 fistyee/mixpro
2	ResNet-200 (DeepAA)	Deep AutoAugment	81.32	2022-03-11	📦 msu-mlsys-lab/deepaa
3	DeiT-S (+MixPro)	MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer	81.30	2023-04-24	📦 fistyee/mixpro
4	ResNet-200 (Fast AA)	Fast AutoAugment	80.60	2019-05-01	📦 kakaobrain/fast-autoaugment 📦 ildoonet/pytorch-randaugment 📦 philip-bachman/amdim-public
5	ResNet-200 (UA)	UniformAugment: A Search-free Probabilistic Data Augmentation Approach	80.40	2020-03-31	📦 tgilewicz/uniformaugment
6	ResNet-200 (AA)	AutoAugment: Learning Augmentation Policies from Data	80.00	2018-05-24	📦 tensorflow/models 📦 tensorflow/models 📦 pytorch/vision
7	ResNet-50 (DeepAA)	Deep AutoAugment	78.30	2022-03-11	📦 msu-mlsys-lab/deepaa
8	ResNet-50 (TA wide)	TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation	78.07	2021-03-18	📦 pytorch/vision 📦 automl/trivialaugment
9	ResNet-50 (LoRot-E)	Tailoring Self-Supervision for Supervised Learning	77.72	2022-07-20	📦 wjun0830/localizable-rotation
10	ResNet-50 (LoRot-I)	Tailoring Self-Supervision for Supervised Learning	77.71	2022-07-20	📦 wjun0830/localizable-rotation

All Papers (17)

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

2023

DeiT-B (+MixPro)

fistyee/mixpro

Deep AutoAugment

2022

ResNet-200 (DeepAA)

msu-mlsys-lab/deepaa

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

2023

DeiT-S (+MixPro)

fistyee/mixpro

Fast AutoAugment

2019

ResNet-200 (Fast AA)

kakaobrain/fast-autoaugment ildoonet/pytorch-randaugment

UniformAugment: A Search-free Probabilistic Data Augmentation Approach

2020

ResNet-200 (UA)

tgilewicz/uniformaugment

AutoAugment: Learning Augmentation Policies from Data

2018

ResNet-200 (AA)

tensorflow/models tensorflow/models

Deep AutoAugment

2022

ResNet-50 (DeepAA)

msu-mlsys-lab/deepaa

TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation

2021

ResNet-50 (TA wide)

pytorch/vision automl/trivialaugment

Tailoring Self-Supervision for Supervised Learning

2022

ResNet-50 (LoRot-E)

wjun0830/localizable-rotation

Tailoring Self-Supervision for Supervised Learning

2022

ResNet-50 (LoRot-I)

wjun0830/localizable-rotation

UniformAugment: A Search-free Probabilistic Data Augmentation Approach

2020

ResNet-50 (UA)

tgilewicz/uniformaugment

AutoAugment: Learning Augmentation Policies from Data

2018

ResNet-50 (AA)

tensorflow/models tensorflow/models

Fast AutoAugment

2019

ResNet-50 (Fast AA)

kakaobrain/fast-autoaugment ildoonet/pytorch-randaugment

RandAugment: Practical automated data augmentation with a reduced search space

2019

ResNet-50 (RA)

rwightman/pytorch-image-models pytorch/vision

DADA: Differentiable Automatic Data Augmentation

2020

ResNet-50 (DADA)

VDIGPKU/DADA

Faster AutoAugment: Learning Augmentation Strategies using Backpropagation

2019

ResNet-50 (Faster AA)

moskomule/dda

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

2023

DeiT-T (+MixPro)

fistyee/mixpro

ImageNet

Performance Over Time

Edit Benchmark Results

Edit Result

Top Performing Models

All Papers (17)

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

Deep AutoAugment

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

Fast AutoAugment

UniformAugment: A Search-free Probabilistic Data Augmentation Approach

AutoAugment: Learning Augmentation Policies from Data

Deep AutoAugment

TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation

Tailoring Self-Supervision for Supervised Learning

Tailoring Self-Supervision for Supervised Learning

UniformAugment: A Search-free Probabilistic Data Augmentation Approach

AutoAugment: Learning Augmentation Policies from Data

Fast AutoAugment

RandAugment: Practical automated data augmentation with a reduced search space

DADA: Differentiable Automatic Data Augmentation

Faster AutoAugment: Learning Augmentation Strategies using Backpropagation

MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

Model	Paper	Accuracy (%)	Date
DeiT-B (+MixPro)	MixPro: Data Augmentation with MaskMix and Progre…	82.90	2023-04-24
ResNet-200 (DeepAA)	Deep AutoAugment	81.32	2022-03-11
DeiT-S (+MixPro)	MixPro: Data Augmentation with MaskMix and Progre…	81.30	2023-04-24
ResNet-200 (Fast AA)	Fast AutoAugment	80.60	2019-05-01
ResNet-200 (UA)	UniformAugment: A Search-free Probabilistic Data …	80.40	2020-03-31
ResNet-200 (AA)	AutoAugment: Learning Augmentation Policies from …	80.00	2018-05-24
ResNet-50 (DeepAA)	Deep AutoAugment	78.30	2022-03-11
ResNet-50 (TA wide)	TrivialAugment: Tuning-free Yet State-of-the-Art …	78.07	2021-03-18
ResNet-50 (LoRot-E)	Tailoring Self-Supervision for Supervised Learning	77.72	2022-07-20
ResNet-50 (LoRot-I)	Tailoring Self-Supervision for Supervised Learning	77.71	2022-07-20
ResNet-50 (UA)	UniformAugment: A Search-free Probabilistic Data …	77.63	2020-03-31
ResNet-50 (AA)	AutoAugment: Learning Augmentation Policies from …	77.60	2018-05-24
ResNet-50 (Fast AA)	Fast AutoAugment	77.60	2019-05-01
ResNet-50 (RA)	RandAugment: Practical automated data augmentatio…	77.60	2019-09-30
ResNet-50 (DADA)	DADA: Differentiable Automatic Data Augmentation	77.50	2020-03-08
ResNet-50 (Faster AA)	Faster AutoAugment: Learning Augmentation Strateg…	76.50	2019-11-16
DeiT-T (+MixPro)	MixPro: Data Augmentation with MaskMix and Progre…	73.80	2023-04-24