Data Augmentation on ImageNet

Metric: Accuracy (%) (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	Accuracy (%)▼	Augmentations	Paper	Date↕	Code
1	DeiT-B (+MixPro)	82.9	No	MixPro: Data Augmentation with MaskMix and Progr...	2023-04-24	Code
2	ResNet-200 (DeepAA)	81.32	No	Deep AutoAugment	2022-03-11	Code
3	DeiT-S (+MixPro)	81.3	No	MixPro: Data Augmentation with MaskMix and Progr...	2023-04-24	Code
4	ResNet-200 (Fast AA)	80.6	No	Fast AutoAugment	2019-05-01	Code
5	ResNet-200 (UA)	80.4	No	UniformAugment: A Search-free Probabilistic Data...	2020-03-31	Code
6	ResNet-200 (AA)	80	No	AutoAugment: Learning Augmentation Policies from...	2018-05-24	Code
7	ResNet-50 (DeepAA)	78.3	No	Deep AutoAugment	2022-03-11	Code
8	ResNet-50 (TA wide)	78.07	No	TrivialAugment: Tuning-free Yet State-of-the-Art...	2021-03-18	Code
9	ResNet-50 (LoRot-E)	77.72	No	Tailoring Self-Supervision for Supervised Learning	2022-07-20	Code
10	ResNet-50 (LoRot-I)	77.71	No	Tailoring Self-Supervision for Supervised Learning	2022-07-20	Code
11	ResNet-50 (UA)	77.63	No	UniformAugment: A Search-free Probabilistic Data...	2020-03-31	Code
12	ResNet-50 (AA)	77.6	No	AutoAugment: Learning Augmentation Policies from...	2018-05-24	Code
13	ResNet-50 (Fast AA)	77.6	No	Fast AutoAugment	2019-05-01	Code
14	ResNet-50 (RA)	77.6	No	RandAugment: Practical automated data augmentati...	2019-09-30	Code
15	ResNet-50 (DADA)	77.5	No	DADA: Differentiable Automatic Data Augmentation	2020-03-08	Code
16	ResNet-50 (Faster AA)	76.5	No	Faster AutoAugment: Learning Augmentation Strate...	2019-11-16	Code
17	DeiT-T (+MixPro)	73.8	No	MixPro: Data Augmentation with MaskMix and Progr...	2023-04-24	Code

#1DeiT-B (+MixPro)SOTA
82.9
Accuracy (%)· 2023-04-24
MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer Code
#2ResNet-200 (DeepAA)SOTA
81.32
Accuracy (%)· 2022-03-11
Deep AutoAugment Code
#3DeiT-S (+MixPro)
81.3
Accuracy (%)· 2023-04-24
MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer Code
#4ResNet-200 (Fast AA)SOTA
80.6
Accuracy (%)· 2019-05-01
Fast AutoAugment Code
#5ResNet-200 (UA)
80.4
Accuracy (%)· 2020-03-31
UniformAugment: A Search-free Probabilistic Data Augmentation Approach Code
#6ResNet-200 (AA)SOTA
80
Accuracy (%)· 2018-05-24
AutoAugment: Learning Augmentation Policies from Data Code
#7ResNet-50 (DeepAA)
78.3
Accuracy (%)· 2022-03-11
Deep AutoAugment Code
#8ResNet-50 (TA wide)
78.07
Accuracy (%)· 2021-03-18
TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation Code
#9ResNet-50 (LoRot-E)
77.72
Accuracy (%)· 2022-07-20
Tailoring Self-Supervision for Supervised Learning Code
#10ResNet-50 (LoRot-I)
77.71
Accuracy (%)· 2022-07-20
Tailoring Self-Supervision for Supervised Learning Code
#11ResNet-50 (UA)
77.63
Accuracy (%)· 2020-03-31
UniformAugment: A Search-free Probabilistic Data Augmentation Approach Code
#12ResNet-50 (AA)
77.6
Accuracy (%)· 2018-05-24
AutoAugment: Learning Augmentation Policies from Data Code
#13ResNet-50 (Fast AA)
77.6
Accuracy (%)· 2019-05-01
Fast AutoAugment Code
#14ResNet-50 (RA)
77.6
Accuracy (%)· 2019-09-30
RandAugment: Practical automated data augmentation with a reduced search space Code
#15ResNet-50 (DADA)
77.5
Accuracy (%)· 2020-03-08
DADA: Differentiable Automatic Data Augmentation Code
#16ResNet-50 (Faster AA)
76.5
Accuracy (%)· 2019-11-16
Faster AutoAugment: Learning Augmentation Strategies using Backpropagation Code
#17DeiT-T (+MixPro)
73.8
Accuracy (%)· 2023-04-24
MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer Code