Domain Generalization on DomainNet (test)
61.8AccuracyMetaPrompt
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MetaPromptFew-shot=5-shot, CLIP=true2022.12 | 61.8 | — | — | — | — | — | — | |
| MetaPromptFew-shot=1-shot, CLIP=true2022.12 | 61.5 | — | — | — | — | — | — | |
| CoCoOpFew-shot=5-shot, CLIP=true2022.12 | 61.1 | — | — | — | — | — | — | |
| EoAEnsemble=true, CLIP=true2022.12 | 60.9 | — | — | — | — | — | — | |
| CLIPZero-shot=true, CLIP=true2022.12 | 57.9 | — | — | — | — | — | — | |
| CLIP (template)Zero-shot=true, CLIP=true2022.12 | 57.8 | — | — | — | — | — | — | |
| SEDGEEnsemble=true, CLIP=true2022.12 | 54.7 | — | — | — | — | — | — | |
| SWAD + OursBackbone=ResNet-50, Protocol=Source free domain adaptation2023.03 | 51.6 | 69.2 | 28.4 | 58.2 | 26.2 | 68.1 | 59.6 | |
| F-mixBackbone=ResNet-50, Protocol=Source free domain adaptation2023.03 | 51 | 75.4 | 24.6 | 57.8 | 23.6 | 65.8 | 58.5 | |
| PCLBackbone=ResNet-50, Protocol=Domain generalization2023.03 | 47.7 | 67.9 | 24.3 | 55.3 | 15.7 | 66.6 | 56.4 | |
| OursBackbone=ResNet-50, Protocol=Source free domain adaptation2023.03 | 47.7 | 66.1 | 24.1 | 52.8 | 18.2 | 68.5 | 56.7 | |
| DNABackbone=ResNet-50, Protocol=Domain generalization2023.03 | 47.2 | 66.1 | 23 | 54.6 | 16.7 | 65.8 | 56.8 | |
| SWADBackbone=ResNet-50, Protocol=Domain generalization2023.03 | 46.7 | 66.1 | 22.4 | 53.6 | 16.3 | 65.5 | 56.2 | |
| SWADEnsemble=true2022.12 | 46.5 | — | — | — | — | — | — | |
| ERMBackbone=ResNet-50, Protocol=Domain generalization2023.03 | 45.2 | 64.8 | 22.1 | 51.8 | 13.8 | 64.7 | 54 | |
| StableFDGBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 44.48 | 62.58 | 24.12 | 52.23 | 14.87 | 60.6 | 52.5 | |
| ADAHESSIANBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 44.4 | — | — | — | — | — | — | |
| FADBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 44.4 | — | — | — | — | — | — | |
| GENIEBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 44.1 | — | — | — | — | — | — | |
| ERM2022.12 | 44 | — | — | — | — | — | — | |
| DSUBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 43.9 | 62.43 | 24.3 | 51.87 | 13.65 | 58.75 | 52.4 | |
| GAMBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.8 | — | — | — | — | — | — | |
| FedSRBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 43.6 | 61.91 | 25.37 | 50.54 | 11.59 | 62.03 | 50.13 | |
| YOGIBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.5 | — | — | — | — | — | — | |
| ADABELIEFBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.5 | — | — | — | — | — | — | |
| ADAMWBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.4 | — | — | — | — | — | — | |
| SAMBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.4 | — | — | — | — | — | — | |
| MixStyleBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 43.25 | 61.39 | 24.33 | 51.41 | 13.07 | 57.9 | 51.4 | |
| SGDBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43.2 | — | — | — | — | — | — | |
| FedAvgBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 43.18 | 61.52 | 24.75 | 50.83 | 12.08 | 60 | 49.92 | |
| ADAMBackbone=ResNet-50, Pre-training=ImageNet, Training Iterations=15,000, Evaluation Protocol=DomainBed (Training-domain Validation Set)2026.06 | 43 | — | — | — | — | — | — | |
| FedDGBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 42.94 | 62.49 | 23.71 | 48.42 | 12.45 | 61.44 | 49.11 | |
| FedBNBackbone=ResNet-50, Global rounds=60, Mini-batch size=64, Scenario=single-domain data distribution2023.11 | 42.71 | 60.23 | 24.4 | 50.43 | 11.99 | 59.57 | 49.66 | |
| Fish2022.12 | 42.7 | — | — | — | — | — | — | |
| MLDG2022.12 | 41.8 | — | — | — | — | — | — | |
| CORAL2022.12 | 41.5 | — | — | — | — | — | — |