Zero-shot Learning on AWA2
0.935Top-1 AccuracyZeroDiff++
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ZeroDiff++Venue=Ours, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.935 | — | — | — | — | — | |
| ZeroDiffVenue=ICLR25, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.873 | — | — | — | — | — | |
| RLVCBackbone=ViT2026.03 | 0.84 | — | — | — | — | — | |
| VADSVenue=CVPR24, Backbone=ViT, Fine-tune features=Other fine-tune features2026.02 | 0.825 | — | — | — | — | — | |
| VADSBackbone=ViT, Method Category=Generative2026.03 | 0.825 | — | — | — | — | — | |
| PSVMA++Venue=TPAMI25, Backbone=ViT, Fine-tune features=None2026.02 | 0.792 | — | — | — | — | — | |
| PSVMA+Backbone=ViT, Method Category=Embedding2026.03 | 0.792 | — | — | — | — | — | |
| ICCEBackbone=ResNet, Method Category=Generative2026.03 | 0.784 | — | — | — | — | — | |
| SC-EGGBackbone=ResNet, Method Category=Generative2026.03 | 0.782 | — | — | — | — | — | |
| ZSL-KGKnowledge Graph=ConceptNet2020.06 | 0.781 | 0.668 | 0.844 | 0.746 | — | — | |
| ViFRBackbone=ResNet, Method Category=Generative2026.03 | 0.778 | — | — | — | — | — | |
| ZSML2020.06 | 0.775 | 0.589 | 0.746 | 0.658 | — | — | |
| I2DFormer+Backbone=ViT, Method Category=Embedding2026.03 | 0.773 | — | — | — | — | — | |
| SGCN2020.06 | 0.771 | 0.675 | 0.812 | 0.737 | — | — | |
| DGP2020.06 | 0.771 | 0.713 | 0.794 | 0.751 | — | — | |
| GCNZ2020.06 | 0.77 | 0.666 | 0.816 | 0.733 | — | — | |
| VSPCNBackbone=ViT, Method Category=Embedding2026.03 | 0.766 | — | — | — | — | — | |
| AGZSL2020.06 | 0.764 | 0.69 | 0.865 | 0.768 | — | — | |
| I2DFormer-WikiVenue=NeurIPS'22, Feature Type=ViT2024.04 | 0.764 | — | — | — | — | — | |
| DiffusionZSLVenue=ICCV23, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.764 | — | — | — | — | — | |
| f-CLSWGANVenue=CVPR18, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.759 | — | — | — | — | — | |
| f-VAEGANVenue=CVPR19, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.758 | — | — | — | — | — | |
| AENetBackbone=ViT, Method Category=Embedding2026.03 | 0.752 | — | — | — | — | — | |
| DFCAFlowVenue=TCSVT23, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.744 | — | — | — | — | — | |
| APN+ABPLearning Paradigm=Generative, Backbone Features=APN features2022.04 | 0.739 | — | — | — | — | — | |
| ViFRCategory=Generative Methods2026.03 | 0.737 | — | — | — | — | — | |
| I2MVFormer-WikiVenue=CVPR'23, Feature Type=ViT2024.04 | 0.736 | — | — | — | — | — | |
| APN+TF-VAEGANLearning Paradigm=Generative, Backbone Features=APN features2022.04 | 0.735 | — | — | — | — | — | |
| APN+f-VAEGAN-D2*Learning Paradigm=Generative, Backbone Features=APN features2022.04 | 0.734 | — | — | — | — | — | |
| TF-VAEGAN*Learning Paradigm=Generative, Backbone Features=ResNet101 (finetuned)2022.04 | 0.734 | — | — | — | — | — | |
| E-PGN2020.06 | 0.734 | 0.526 | 0.835 | 0.646 | — | — | |
| TF-VAEGAN2020.06 | 0.734 | 0.555 | 0.836 | 0.667 | — | — | |
| MSDN++Category=Embedding-based Methods, Attention-based=true2026.03 | 0.734 | — | — | — | — | — | |
| ZeroNASBackbone=ResNet, Method Category=Generative2026.03 | 0.732 | — | — | — | — | — | |
| CEGANVenue=CVPR21, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.728 | — | — | — | — | — | |
| TransZero++2020.06 | 0.726 | 0.646 | 0.827 | 0.725 | — | — | |
| TransZero++Venue=TPAMI22, Backbone=Res101, Fine-tune features=None2026.02 | 0.726 | — | — | — | — | — | |
| TransZero++Backbone=ResNet, Method Category=Embedding2026.03 | 0.726 | — | — | — | — | — | |
| TFVAEGANVenue=ECCV20, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.724 | — | — | — | — | — | |
| TF-VAEGAN2021.03 | 0.722 | — | — | — | — | — | |
| TF-VAEGANVenue=ECCV'20, Feature Type=CNN2024.04 | 0.722 | — | — | — | — | — | |
| ZeroMambaBackbone=VMamba, Method Category=Embedding2026.03 | 0.719 | — | — | — | — | — | |
| ERPCNet2020.06 | 0.718 | 0.591 | 0.82 | 0.687 | — | — | |
| APN2020.06 | 0.717 | 0.622 | 0.695 | 0.656 | — | — | |
| ComposerVenue=NeurIPS'20, Feature Type=CNN, Attention=true2024.04 | 0.715 | — | — | — | — | — | |
| ComposerCategory=Generative Methods, Attention-based=true2026.03 | 0.715 | — | — | — | — | — | |
| HASBackbone=ResNet, Method Category=Embedding2026.03 | 0.714 | — | — | — | — | — | |
| OCD-CVAE2020.06 | 0.713 | 0.595 | 0.734 | 0.657 | — | — | |
| TCN2021.03 | 0.712 | — | — | — | — | — | |
| APN+CVCLearning Paradigm=Generative, Backbone Features=APN features2022.04 | 0.712 | — | — | — | — | — | |
| f-VAEGAN-D22021.03 | 0.711 | — | — | — | — | — | |
| GXE2020.06 | 0.711 | 0.564 | 0.814 | 0.667 | — | — | |
| f-VAEGANVenue=CVPR'19, Feature Type=CNN2024.04 | 0.711 | — | — | — | — | — | |
| f-VAEGAN-D2Category=Generative Methods2026.03 | 0.711 | — | — | — | — | — | |
| GNDANCategory=Embedding-based Methods, Attention-based=true2026.03 | 0.71 | — | — | — | — | — | |
| ReZSLVenue=TIP23, Backbone=Res101, Fine-tune features=None2026.02 | 0.709 | — | — | — | — | — | |
| ZSLViTFeature Type=ViT2024.04 | 0.707 | — | — | — | — | — | |
| ZSLViTVenue=CVPR24, Backbone=ViT, Fine-tune features=None2026.02 | 0.707 | — | — | — | — | — | |
| ZSLViTBackbone=ViT, Method Category=Embedding2026.03 | 0.707 | — | — | — | — | — | |
| LisGANLearning Paradigm=Generative2022.04 | 0.706 | — | — | — | — | — | |
| EGANSBackbone=ResNet, Method Category=Generative2026.03 | 0.706 | — | — | — | — | — | |
| Zhu et al.2021.03 | 0.704 | — | — | — | — | — | |
| CE-GZSL2021.03 | 0.704 | — | — | — | — | — | |
| CE-GZSL2020.06 | 0.704 | 0.631 | 0.786 | 0.7 | — | — | |
| CE-GZSLBackbone=ResNet, Method Category=Generative2026.03 | 0.704 | — | — | — | — | — | |
| DLFZRL2021.03 | 0.703 | — | — | — | — | — | |
| f-VAEGAN-D2*Learning Paradigm=Generative, Backbone Features=ResNet101 (finetuned)2022.04 | 0.703 | — | — | — | — | — | |
| f-VAEGAN-D22020.06 | 0.703 | 0.571 | 0.761 | 0.652 | — | — | |
| TransZero2020.06 | 0.701 | 0.613 | 0.823 | 0.702 | — | — | |
| MSDN2020.06 | 0.701 | 0.62 | 0.745 | 0.677 | — | — | |
| MSDNVenue=CVPR'22, Feature Type=CNN, Attention=true2024.04 | 0.701 | — | — | — | — | — | |
| TransZeroVenue=AAAI'22, Feature Type=CNN, Attention=true2024.04 | 0.701 | — | — | — | — | — | |
| TransZeroCategory=Embedding-based Methods, Attention-based=true2026.03 | 0.701 | — | — | — | — | — | |
| MSDNCategory=Embedding-based Methods, Attention-based=true2026.03 | 0.701 | — | — | — | — | — | |
| APNLearning Paradigm=Non-generative, ZoomInMod=true2022.04 | 0.699 | — | — | — | — | — | |
| DUETVenue=AAAI'23, Feature Type=ViT2024.04 | 0.699 | — | — | — | — | — | |
| APNCategory=Embedding-based Methods, Attention-based=true2026.03 | 0.699 | — | — | — | — | — | |
| DUETBackbone=ViT, Method Category=Embedding2026.03 | 0.699 | — | — | — | — | — | |
| SDVAEVenue=ICCV21, Backbone=Res101, Fine-tune features=Author's fine-tune features2026.02 | 0.693 | — | — | — | — | — | |
| TAFE-NetAdditional synthetic data=false2019.04 | 0.693 | — | — | — | — | — | |
| SEAdditional synthetic data=true2019.04 | 0.692 | — | — | — | — | — | |
| SGMALearning Paradigm=Non-generative2022.04 | 0.688 | — | — | — | — | — | |
| SGMACategory=Embedding-based Methods, Attention-based=true2026.03 | 0.688 | — | — | — | — | — | |
| ABP*Learning Paradigm=Generative, Backbone Features=ResNet101 (finetuned)2022.04 | 0.685 | — | — | — | — | — | |
| APN+Learning Paradigm=Non-generative, ZoomInMod=false2022.04 | 0.684 | — | — | — | — | — | |
| RSR2020.06 | 0.684 | 0.553 | 0.76 | 0.64 | — | — | |
| APNVenue=NeurIPS'20, Feature Type=CNN, Attention=true2024.04 | 0.684 | — | — | — | — | — | |
| APNVenue=NeurIPS20, Backbone=Res101, Fine-tune features=None2026.02 | 0.684 | — | — | — | — | — | |
| CLSWGANLearning Paradigm=Generative2022.04 | 0.682 | — | — | — | — | — | |
| f-CLSWGANCategory=Generative Methods2026.03 | 0.682 | — | — | — | — | — | |
| LFGAA2021.03 | 0.681 | — | — | — | — | — | |
| LFGAA+HybridLearning Paradigm=Non-generative2022.04 | 0.681 | — | — | — | — | — | |
| LFGAA+Hibrid2020.06 | 0.681 | 0.27 | 0.934 | 0.419 | — | — | |
| LFGAACategory=Embedding-based Methods, Attention-based=true2026.03 | 0.681 | — | — | — | — | — | |
| ARENLearning Paradigm=Non-generative2022.04 | 0.679 | — | — | — | — | — | |
| ARENVenue=CVPR'19, Feature Type=CNN, Attention=true2024.04 | 0.679 | — | — | — | — | — | |
| DAZLEVenue=CVPR'20, Feature Type=CNN, Attention=true2024.04 | 0.679 | — | — | — | — | — | |
| ARENCategory=Embedding-based Methods, Attention-based=true2026.03 | 0.679 | — | — | — | — | — | |
| DAZLECategory=Embedding-based Methods, Attention-based=true2026.03 | 0.679 | — | — | — | — | — | |
| GEM-ZSLVenue=CVPR'22, Feature Type=CNN, Attention=true2024.04 | 0.673 | — | — | — | — | — |