Domain Generalization On Imagenet A

Metrics

Top-1 accuracy %

Results

Performance results of various models on this benchmark

Model Name	Top-1 accuracy %	Paper Title	Repository
Discrete Adversarial Distillation (ResNet-50)	7.7	Distilling Out-of-Distribution Robustness from Vision-Language Foundation Models	-
CutMix (ResNet-50)	7.3	CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features	-
RVT-Ti*	14.4	Towards Robust Vision Transformer	-
RVT-S*	25.7	Towards Robust Vision Transformer	-
Cutout (ResNet-50)	4.4	Improved Regularization of Convolutional Neural Networks with Cutout	-
ResNet-50	0	Natural Adversarial Examples	-
µ2Net+ (ViT-L/16)	84.53	A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems	-
CAFormer-B36 (IN-21K)	69.4	MetaFormer Baselines for Vision	-
MAE (ViT-H, 448)	76.7	Masked Autoencoders Are Scalable Vision Learners	-
CAFormer-B36	48.5	MetaFormer Baselines for Vision	-
FAN-Hybrid-L(IN-21K, 384)	74.5	Understanding The Robustness in Vision Transformers	-
ResNet-50 (300 Epochs)	4.2	Deep Residual Learning for Image Recognition	-
Model soups (BASIC-L)	94.17	Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time	-
FAN-L-Hybrid+STL	46.1	Fully Attentional Networks with Self-emerging Token Labeling	-
TransNeXt-Base (IN-1K supervised, 224)	50.6	TransNeXt: Robust Foveal Visual Perception for Vision Transformers	-
CAFormer-B36 (IN-21K, 384)	79.5	MetaFormer Baselines for Vision	-
Model soups (ViT-G/14)	92.67	Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time	-
CutMix+MoEx (ResNet-50)	8.4	On Feature Normalization and Data Augmentation	-
Stylized ImageNet (ResNet-50)	2.3	ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness	-
ConvFormer-B36 (384)	55.3	MetaFormer Baselines for Vision	-

0 of 39 row(s) selected.