AugMix (ResNet-50)

Reported on 4 benchmarks across 2 tasks · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology2 results

Domain AdaptationonImageNet-C
mean Corruption Error (mCE)· 2019-12-05
65.3
best: 22 (EfficientNet-L2+RPL)
SOTA
AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty arXiv:1912.02781
Domain AdaptationonImageNet-R
Top-1 Error Rate· 2019-12-05
58.9
best: 3.9 (Model soups (BASIC-L))
AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty arXiv:1912.02781

Computer Vision2 results

Domain GeneralizationonImageNet-C
mean Corruption Error (mCE)· 2019-12-05
65.3
best: 28.2 (DINOv2 (ViT-g/14, frozen model, linear eval))
SOTA
AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty arXiv:1912.02781
Domain GeneralizationonImageNet-R
Top-1 Error Rate· 2019-12-05
58.9
best: 3.9 (Model soups (BASIC-L))
AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty arXiv:1912.02781