Classification on DTD
88.6AccuracyDREAM
Evaluation Results
| Method | Links | |
|---|---|---|
| DREAMWay=5, Shot=52026.03 | 88.6 | |
| CLIPWay=5, Shot=52026.03 | 81 | |
| true-baseEpoch Selection=Last Epoch2024.05 | 78.1 | |
| true-baseEpoch Selection=Best Epoch2024.05 | 77.6 | |
| REPAWay=5, Shot=52026.03 | 73.2 | |
| SigLIP2Data size=10B, Zero-shot=true2026.03 | 67.2 | |
| PEcoreData size=5B, Zero-shot=true2026.03 | 66.4 | |
| FLUIDWay=5, Shot=52026.03 | 66.3 | |
| MARWay=5, Shot=52026.03 | 64.9 | |
| SigLIPData size=10B, Zero-shot=true2026.03 | 64.7 | |
| LlipData size=2.5B, Zero-shot=true2026.03 | 63.7 | |
| MetaCLIPData size=2.5B, Zero-shot=true2026.03 | 62.1 | |
| ITOPre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 61.4 | |
| ITO sub2Pre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 60.4 | |
| CLIPPre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 59.4 | |
| annot-mixEpoch Selection=Last Epoch2024.05 | 55.2 | |
| annot-mixEpoch Selection=Best Epoch2024.05 | 55.1 | |
| OpenCLIPData size=2B, Zero-shot=true2026.03 | 54.9 | |
| BiFTAZero-shot=true, Backbone=Average over five CLIP variants2026.01 | 54.52 | |
| BiFTABackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 53.22 | |
| WCAZero-shot=true, Backbone=Average over five CLIP variants2026.01 | 52.77 | |
| BiFTABackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 51.41 | |
| WCABackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 49.89 | |
| WCABackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 49.45 | |
| madlEpoch Selection=Best Epoch2024.05 | 47.7 | |
| CuPLBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 47.5 | |
| CuPLBackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 47.39 | |
| madlEpoch Selection=Last Epoch2024.05 | 47.2 | |
| mv-baseEpoch Selection=Best Epoch2024.05 | 46.9 | |
| trace-regEpoch Selection=Best Epoch2024.05 | 46.9 | |
| mv-mixupEpoch Selection=Best Epoch2024.05 | 46.2 | |
| ds-mixupEpoch Selection=Best Epoch2024.05 | 46.2 | |
| CLIP-DBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 44.2 | |
| CLIP-EBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 43.83 | |
| mv-mixupEpoch Selection=Last Epoch2024.05 | 43 | |
| ds-mixupEpoch Selection=Last Epoch2024.05 | 43 | |
| WaffleBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 42.98 | |
| CLIPBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.01 | 42.93 | |
| CLIP-DBackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 42.23 | |
| USEBackbone=ResNet50, Optim.=true2026.07 | 41.84 | |
| conalEpoch Selection=Best Epoch2024.05 | 41.8 | |
| RLCFBackbone=ResNet50, Optim.=true2026.07 | 41.61 | |
| TPTBackbone=ResNet50, Optim.=true2026.07 | 41.55 | |
| C-TPTBackbone=ResNet50, Optim.=true2026.07 | 41.34 | |
| R-TPTBackbone=ResNet50, Optim.=true2026.07 | 41.31 | |
| CLIP-EBackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 41.22 | |
| conalEpoch Selection=Last Epoch2024.05 | 40.9 | |
| SEBackbone=ResNet50, Optim.=false2026.07 | 40.54 | |
| CLIPBackbone=ResNet50, Optim.=false2026.07 | 40.37 | |
| TPSBackbone=ResNet50, Optim.=true2026.07 | 40.25 | |
| crowd-arEpoch Selection=Best Epoch2024.05 | 40.1 | |
| MTABackbone=ResNet50, Optim.=false2026.07 | 39.92 | |
| ZEROBackbone=ResNet50, Optim.=false2026.07 | 39.86 | |
| STSBackbone=ResNet50, Optim.=true2026.07 | 39.63 | |
| crowd-arEpoch Selection=Last Epoch2024.05 | 39.3 | |
| WaffleBackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 38.88 | |
| CLIPBackbone=RN-50, Zero-shot evaluation protocol=true2026.01 | 38.67 | |
| FLAIRData size=30M, Zero-shot=true2026.03 | 37.3 | |
| COSMOSData size=30M, Zero-shot=true2026.03 | 37.3 | |
| trace-regEpoch Selection=Last Epoch2024.05 | 36.6 | |
| geo-reg-fEpoch Selection=Best Epoch2024.05 | 35.9 | |
| geo-reg-wEpoch Selection=Best Epoch2024.05 | 35.8 | |
| mv-baseEpoch Selection=Last Epoch2024.05 | 35.6 | |
| geo-reg-fEpoch Selection=Last Epoch2024.05 | 35.2 | |
| geo-reg-wEpoch Selection=Last Epoch2024.05 | 34.6 | |
| DreamLIPData size=30M, Zero-shot=true2026.03 | 33.9 | |
| GoldiCLIPData size=30M, Zero-shot=true2026.03 | 33.9 | |
| union-netEpoch Selection=Best Epoch2024.05 | 33.6 | |
| SigLIPData size=30M, Zero-shot=true2026.03 | 33.5 | |
| crowd-layerEpoch Selection=Best Epoch2024.05 | 31.9 | |
| crowd-layerEpoch Selection=Last Epoch2024.05 | 31.3 | |
| CLIPData size=30M, Zero-shot=true2026.03 | 30.9 | |
| union-netEpoch Selection=Last Epoch2024.05 | 30.9 | |
| FLVMIAAlignment=frozen pretrained DINOv2 and GTE-1.5, Setting=Zero-shot2026.05 | 21.6 | |
| FLQMIAAlignment=frozen pretrained DINOv2 and GTE-1.5, Setting=Zero-shot2026.05 | 21.44 | |
| ITO sub3Backbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 21.3 | |
| MetaMAEPretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 20.9 | |
| ITO sub2Backbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 20.5 | |
| FLAIRBackbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 18.8 | |
| ShEDPretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 18.4 | |
| CSAAlignment=frozen pretrained DINOv2 and GTE-1.5, Setting=Zero-shot2026.05 | 17.29 | |
| SAIL w/ SigLIPAlignment=frozen pretrained DINOv2 and GTE-1.5, Loss=SigLIP, Setting=Zero-shot2026.05 | 17.18 | |
| CapriPretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 14.7 | |
| MAEPretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 14.2 | |
| SAIL w/ CLIPAlignment=frozen pretrained DINOv2 and GTE-1.5, Loss=InfoNCE, Setting=Zero-shot2026.05 | 13.99 | |
| e-MixPretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 8.8 | |
| BaselinePretrain data=ImageNet32, Protocol=Linear evaluation2023.10 | 7.4 |