Few-shot classification on CUB
93.18AccuracyMCL-Katz
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MCL-KatzBackbone=ResNet-12, Shot=5-shot2021.06 | 93.18 | — | — | — | — | — | |
| FRNBackbone=ResNet-12, Shot=5-shot2021.06 | 92.59 | — | — | — | — | — | |
| DN4Backbone=ResNet-12, Shot=5-shot2021.06 | 92.51 | — | — | — | — | — | |
| MCLBackbone=ResNet-12, Shot=5-shot2021.06 | 92.18 | — | — | — | — | — | |
| ProtoNet + MCLBackbone=ResNet-12, Shot=5-shot2021.06 | 91.71 | — | — | — | — | — | |
| DeepEMDBackbone=ResNet-12, Shot=5-shot2021.06 | 91.6 | — | — | — | — | — | |
| ProtoNetBackbone=ResNet-12, Shot=5-shot2021.06 | 91.38 | — | — | — | — | — | |
| DSNBackbone=ResNet-12, Shot=5-shot2021.06 | 91.19 | — | — | — | — | — | |
| CTXBackbone=ResNet-12, Shot=5-shot2021.06 | 90.9 | — | — | — | — | — | |
| MCL-KatzBackbone=Conv-4, Shot=5-shot2021.06 | 90.56 | — | — | — | — | — | |
| SupervisedBackbone=ResNet-50, Evaluation protocol=5-way 5-shot2023.03 | 89.13 | — | — | — | — | — | |
| MCLBackbone=Conv-4, Shot=5-shot2021.06 | 88.71 | — | — | — | — | — | |
| DN4Backbone=Conv-4, Shot=5-shot2021.06 | 88.43 | — | — | — | — | — | |
| FRNBackbone=Conv-4, Shot=5-shot2021.06 | 88.43 | — | — | — | — | — | |
| CTXBackbone=Conv-4, Shot=5-shot2021.06 | 87.31 | — | — | — | — | — | |
| APN+f-VAEGAN-D2Shots=102022.04 | 87.1 | — | — | — | — | — | |
| ProtoNet + MCLBackbone=Conv-4, Shot=5-shot2021.06 | 86.95 | — | — | — | — | — | |
| APN+TF-VAEGANShots=102022.04 | 86.6 | — | — | — | — | — | |
| f-VAEGAN-D2Shots=102022.04 | 85.9 | — | — | — | — | — | |
| DeepEMDBackbone=Conv-4, Shot=5-shot2021.06 | 85.68 | — | — | — | — | — | |
| MCL-KatzBackbone=ResNet-12, Shot=1-shot2021.06 | 85.63 | — | — | — | — | — | |
| TF-VAEGANShots=102022.04 | 85.6 | — | — | — | — | — | |
| DN4Backbone=ResNet-12, Shot=1-shot2021.06 | 85.44 | — | — | — | — | — | |
| DSNBackbone=Conv-4, Shot=5-shot2021.06 | 85.41 | — | — | — | — | — | |
| ImprintedShots=102022.04 | 85.3 | — | — | — | — | — | |
| APN+TF-VAEGANShots=52022.04 | 85.2 | — | — | — | — | — | |
| APN+f-VAEGAN-D2Shots=52022.04 | 84.8 | — | — | — | — | — | |
| ProtoNet + MCLBackbone=ResNet-12, Shot=1-shot2021.06 | 84.59 | — | — | — | — | — | |
| MCLBackbone=ResNet-12, Shot=1-shot2021.06 | 83.64 | — | — | — | — | — | |
| TF-VAEGANShots=52022.04 | 83.5 | — | — | — | — | — | |
| f-VAEGAN-D2Shots=52022.04 | 83.4 | — | — | — | — | — | |
| DeepEMDBackbone=ResNet-12, Shot=1-shot2021.06 | 83.35 | — | — | — | — | — | |
| FRNBackbone=ResNet-12, Shot=1-shot2021.06 | 83.16 | — | — | — | — | — | |
| PsCoBackbone=ResNet-50, Base Method=BYOL, Evaluation protocol=5-way 5-shot2023.03 | 83.13 | — | — | — | — | — | |
| ProtoNetBackbone=ResNet-12, Shot=1-shot2021.06 | 82.98 | — | — | — | — | — | |
| APN+TF-VAEGANShots=22022.04 | 82.6 | — | — | — | — | — | |
| C-HyperbolicShot=5, Way=5, Embedding Space=Clipped Hyperbolic2021.07 | 81.76 | — | — | — | — | — | |
| ProtoNetBackbone=Conv-4, Shot=5-shot2021.06 | 81.5 | — | — | — | — | — | |
| AnalogyShots=102022.04 | 81.1 | — | — | — | — | — | |
| APN+f-VAEGAN-D2Shots=22022.04 | 81.1 | — | — | — | — | — | |
| TF-VAEGANShots=22022.04 | 81.1 | — | — | — | — | — | |
| DSNBackbone=ResNet-12, Shot=1-shot2021.06 | 80.8 | — | — | — | — | — | |
| ImprintedShots=52022.04 | 80 | — | — | — | — | — | |
| MCL-KatzBackbone=Conv-4, Shot=1-shot2021.06 | 79.61 | — | — | — | — | — | |
| f-VAEGAN-D2Shots=22022.04 | 79.6 | — | — | — | — | — | |
| HyperbolicShot=5, Way=5, Embedding Space=Hyperbolic2021.07 | 79.51 | — | — | — | — | — | |
| CTXBackbone=ResNet-12, Shot=1-shot2021.06 | 78.47 | — | — | — | — | — | |
| DN4Backbone=Conv-4, Shot=1-shot2021.06 | 78.31 | — | — | — | — | — | |
| AnalogyShots=52022.04 | 78 | — | — | — | — | — | |
| MCLBackbone=Conv-4, Shot=1-shot2021.06 | 77.8 | — | — | — | — | — | |
| APN+f-VAEGAN-D2Shots=12022.04 | 77.8 | — | — | — | — | — | |
| APN+TF-VAEGANShots=12022.04 | 77.1 | — | — | — | — | — | |
| ProtoNet + MCLBackbone=Conv-4, Shot=1-shot2021.06 | 76.87 | — | — | — | — | — | |
| PsCoBackbone=ResNet-50, Base Method=MoCo v2, Evaluation protocol=5-way 5-shot2023.03 | 76.63 | — | — | — | — | — | |
| f-VAEGAN-D2Shots=12022.04 | 76.1 | — | — | — | — | — | |
| TF-VAEGANShots=12022.04 | 75.6 | — | — | — | — | — | |
| DeepEMDBackbone=Conv-4, Shot=1-shot2021.06 | 75.34 | — | — | — | — | — | |
| FRNBackbone=Conv-4, Shot=1-shot2021.06 | 73.48 | — | — | — | — | — | |
| ProtoNetBackbone=Conv-4, Shot=1-shot2021.06 | 71.64 | — | — | — | — | — | |
| EuclideanShot=5, Way=5, Embedding Space=Euclidean2021.07 | 70.77 | — | — | — | — | — | |
| PsCoBackbone=ResNet-18, Base Method=MoCo v2, Evaluation protocol=5-way 5-shot2023.03 | 70.08 | — | — | — | — | — | |
| CTXBackbone=Conv-4, Shot=1-shot2021.06 | 69.64 | — | — | — | — | — | |
| AnalogyShots=22022.04 | 69.5 | — | — | — | — | — | |
| PsCoway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 68.58 | — | — | — | — | — | |
| BYOLBackbone=ResNet-50, Evaluation protocol=5-way 5-shot2023.03 | 67.45 | — | — | — | — | — | |
| GNN (FT)Backbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=true2020.01 | 66.98 | — | — | — | — | — | |
| DSNBackbone=Conv-4, Shot=1-shot2021.06 | 66.01 | — | — | — | — | — | |
| ProtoNetsway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 65.03 | — | — | — | — | — | |
| ImprintedShots=22022.04 | 65 | — | — | — | — | — | |
| C-HyperbolicShot=1, Way=5, Embedding Space=Clipped Hyperbolic2021.07 | 64.66 | — | — | — | — | — | |
| MAMLway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 64.17 | — | — | — | — | — | |
| MoCo v2Backbone=ResNet-50, Evaluation protocol=5-way 5-shot2023.03 | 64.16 | — | — | — | — | — | |
| MoCo v2way=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 62.81 | — | — | — | — | — | |
| GNNBackbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=false2020.01 | 62.25 | — | — | — | — | — | |
| SimCLRway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 61.89 | — | — | — | — | — | |
| MoCo v2Backbone=ResNet-18, Evaluation protocol=5-way 5-shot2023.03 | 61.88 | — | — | — | — | — | |
| SWAVway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 61.38 | — | — | — | — | — | |
| HyperbolicShot=1, Way=5, Embedding Space=Hyperbolic2021.07 | 61.18 | — | — | — | — | — | |
| RelationNet (FT)Backbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=true2020.01 | 59.46 | — | — | — | — | — | |
| RelationNetBackbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=false2020.01 | 57.77 | — | — | — | — | — | |
| PsCoway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 57.38 | — | — | — | — | — | |
| ProtoNetsway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 56.74 | — | — | — | — | — | |
| MAMLway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 56.57 | — | — | — | — | — | |
| AnalogyShots=12022.04 | 56.5 | — | — | — | — | — | |
| MatchingNet (FT)Backbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=true2020.01 | 55.23 | — | — | — | — | — | |
| MetaOptNet-SVM-trainvalBackbone=ResNet-12, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=false2020.01 | 54.67 | — | — | — | — | — | |
| Meta-SVEBMway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 54.61 | — | — | — | — | — | |
| Meta-GMVAEway=5, shot=20, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 54.08 | — | — | — | — | — | |
| MoCo v2way=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 53.23 | — | — | — | — | — | |
| SimCLRway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 52.11 | — | — | — | — | — | |
| SWAVway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 51.58 | — | — | — | — | — | |
| MatchingNetBackbone=ResNet-10, Training Dataset=mini-ImageNet, FT (Feature-wise Transformation)=false2020.01 | 51.37 | — | — | — | — | — | |
| EuclideanShot=1, Way=5, Embedding Space=Euclidean2021.07 | 51.31 | — | — | — | — | — | |
| ImprintedShots=12022.04 | 48.5 | — | — | — | — | — | |
| Meta-GMVAEway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 47.48 | — | — | — | — | — | |
| Meta-SVEBMway=5, shot=5, Backbone=Conv5, Pre-trained=miniImageNet2023.03 | 45.5 | — | — | — | — | — | |
| Oursway=5, shot=5, scenario=multi-domain2026.07 | 44.8 | — | — | — | — | — | |
| PsCoway=5, shot=5, scenario=multi-domain2026.07 | 36.1 | — | — | — | — | — | |
| KD-MAMLway=5, shot=5, scenario=multi-domain2026.07 | 35.1 | — | — | — | — | — | |
| Oursway=5, shot=1, scenario=multi-domain2026.07 | 34.7 | — | — | — | — | — |