5-way Few-shot Classification on tiered-ImageNet
86.03AccuracyDeepEMD
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DeepEMDBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 86.03 | — | — | |
| FEAT-Res12+1S-DAug-2Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 85.55 | — | — | |
| DFRBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 85.12 | — | — | |
| FEAT-Res12+1S-DAug-1Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 84.82 | — | — | |
| FEAT-Res12Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 84.79 | — | — | |
| ProtoNet-Res12Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 84.03 | — | — | |
| ProtoNet-Res12+1S-DAug-1Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 83.86 | — | — | |
| Meta-BaselineBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 83.74 | — | — | |
| Classifier-BaselineBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 83.74 | — | — | |
| FEAT-Res12 (re-impl.)Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 82.21 | — | — | |
| MetaOptNetShot=5-shot, Network Architecture=non-standard network2020.03 | 81.75 | — | — | |
| MetaOptNetBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 81.56 | — | — | |
| LEOShot=5-shot, Network Architecture=non-standard network2020.03 | 81.44 | — | — | |
| ProtoNet-Res12 (re-impl.)Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 81.13 | — | — | |
| SImPaShot=5-shot, Network Architecture=non-standard network, Input Features=extracted features [15]2020.03 | 80.15 | — | — | |
| FEAT-Res12+1S-DAug-2Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 73.18 | — | — | |
| Proto. NetsShot=5-shot, Network Architecture=non-standard network2020.03 | 72.69 | — | — | |
| FEAT-Res12+1S-DAug-1Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 71.85 | — | — | |
| Trns. Prp. NetsShot=5-shot, Network Architecture=non-standard network2020.03 | 71.55 | — | — | |
| Relation NetShot=5-shot, Network Architecture=non-standard network2020.03 | 71.32 | — | — | |
| DFRBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 71.31 | — | — | |
| DeepEMDBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 71.16 | — | — | |
| FEAT-Res12Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 70.8 | — | — | |
| ProtoNet-Res12+1S-DAug-2Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 70.32 | — | — | |
| MAMLShot=5-shot, Network Architecture=non-standard network2020.03 | 70.3 | — | — | |
| SImPaShot=1-shot, Network Architecture=non-standard network, Input Features=extracted features [15]2020.03 | 70.26 | — | — | |
| ProtoNet-Res12+1S-DAug-1Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 69.06 | — | — | |
| Meta-BaselineBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 68.62 | — | — | |
| FEAT-Res12 (re-impl.)Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 68.28 | — | — | |
| ProtoNet-Res12Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 68.23 | — | — | |
| Classifier-BaselineBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 68.07 | — | — | |
| LEOShot=1-shot, Network Architecture=non-standard network2020.03 | 66.33 | — | — | |
| MetaOptNetBackbone=Res12, Evaluation Protocol=Inductive2026.01 | 65.99 | — | — | |
| MetaOptNetShot=1-shot, Network Architecture=non-standard network2020.03 | 65.81 | — | — | |
| ProtoNet-Res12 (re-impl.)Backbone=Res12, Evaluation Protocol=Inductive2026.01 | 65.28 | — | — | |
| Trns. Prp. NetsShot=1-shot, Network Architecture=non-standard network2020.03 | 57.41 | — | — | |
| Relation NetShot=1-shot, Network Architecture=non-standard network2020.03 | 54.48 | — | — | |
| Proto. NetsShot=1-shot, Network Architecture=non-standard network2020.03 | 53.31 | — | — | |
| MAMLShot=1-shot, Network Architecture=non-standard network2020.03 | 51.67 | — | — | |
| BD-CSPNBackbone=ResNet-182021.06 | — | 77.28 | 86.55 | |
| BD-CSPNBackbone=WRN2021.06 | — | 78.89 | 87.72 | |
| BD-CSPN + ESFRBackbone=ResNet-182021.06 | — | 80.13 | 86.34 | |
| BD-CSPN + ESFRBackbone=WRN2021.06 | — | 81.77 | 87.61 | |
| BD-CSPN + ESFR-SemiBackbone=ResNet-182021.06 | — | — | 86.83 | |
| BD-CSPN + ESFR-SemiBackbone=WRN2021.06 | — | — | 88.1 | |
| CTMBackbone=ResNet-182021.06 | — | 68.41 | 84.28 | |
| DeepEMDBackbone=ResNet122022.10 | — | 71.16 | 86.03 | |
| E3BMBackbone=ResNet252022.10 | — | 70 | 85 | |
| FEATBackbone=ResNet-182021.06 | — | 70.8 | 84.79 | |
| FEATBackbone=WRN2021.06 | — | 70.41 | 84.38 | |
| Free-LunchBackbone=ResNet12, Implementation=H-OT paper reproduction2022.10 | — | 71.4 | 85.56 | |
| Free-LunchBackbone=WRN28+RTloss2022.10 | — | 75.1 | 88.42 | |
| H-OTBackbone=ResNet122022.10 | — | 73.71 | 87.46 | |
| H-OTBackbone=WRN28+RTloss2022.10 | — | 75.91 | 89.33 | |
| LaplacianShotBackbone=ResNet-182021.06 | — | 78.98 | 86.39 | |
| LaplacianShotBackbone=WRN2021.06 | — | 80.18 | 87.56 | |
| LEOBackbone=WRN2021.06 | — | 66.33 | 81.44 | |
| LEOBackbone=WRN282022.10 | — | 66.33 | 81.44 | |
| MatchNetBackbone=ResNet122022.10 | — | 68.5 | 80.6 | |
| ProtoNetBackbone=ResNet122022.10 | — | 65.65 | 83.4 | |
| SimpleShotBackbone=ResNet-182021.06 | — | 69.68 | 84.56 | |
| SimpleShotBackbone=WRN2021.06 | — | 70.9 | 85.76 | |
| Tran. BaselineBackbone=WRN2021.06 | — | 73.34 | 85.5 | |
| WDAE-GNNBackbone=WRN2021.06 | — | 68.18 | 83.09 |