Few-shot Classification on CUB (test)
97.4AccuracyMetaQDA
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MetaQDAWays=5, Shots=5, Setting=Universal Meta-Learning2023.10 | 97.4 | — | — | — | — | |
| MetaOptWays=5, Shots=5, Setting=Universal Meta-Learning, Backbone Status=Frozen2023.10 | 97.2 | — | — | — | — | |
| ProtoNetWays=5, Shots=5, Setting=Universal Meta-Learning, Backbone Status=Frozen2023.10 | 97.1 | — | — | — | — | |
| CAMLWays=5, Shots=5, Setting=Universal Meta-Learning2023.10 | 97.1 | — | — | — | — | |
| P>M>FWays=5, Shots=5, Setting=In-Domain [Meta-Training], Backbone=ViT-base2023.10 | 97 | — | — | — | — | |
| CausalFSFGPublished In=-, Backbone=ResNet-12, Shot=5-shot2025.12 | 95.26 | — | — | — | — | |
| Bi-FRNPublished In=AAAI 2023, Backbone=ResNet-12, Shot=5-shot2025.12 | 94.73 | — | — | — | — | |
| noHub-SFeature Extractor=WideRes28-10, Shot=5-shot, Classifier=SIAMESE2023.03 | 94.69 | — | — | — | — | |
| noHubFeature Extractor=WideRes28-10, Shot=5-shot, Classifier=SIAMESE2023.03 | 94.56 | — | — | — | — | |
| GPICLWays=5, Shots=5, Setting=Universal Meta-Learning2023.10 | 94.5 | — | — | — | — | |
| FRN+TDMPublished In=CVPR 2022, Backbone=ResNet-12, Shot=5-shot2025.12 | 93.99 | — | — | — | — | |
| FRNPublished In=CVPR 2021, Backbone=ResNet-12, Shot=5-shot2025.12 | 93.34 | — | — | — | — | |
| SNAILWays=5, Shots=5, Setting=Universal Meta-Learning2023.10 | 92.8 | — | — | — | — | |
| P>M>FWays=5, Shots=1, Setting=In-Domain [Meta-Training], Backbone=ViT-base2023.10 | 92.3 | — | — | — | — | |
| ProtoNetPublished In=NeurIPS 2017, Backbone=ResNet-12, Shot=5-shot2025.12 | 91.93 | — | — | — | — | |
| CAMLWays=5, Shots=1, Setting=Universal Meta-Learning2023.10 | 91.8 | — | — | — | — | |
| Variational Feature DisentanglingBackbone=ResNet12, Shots=5, Ways=52020.10 | 91.48 | — | — | — | — | |
| TEENWay=5-way, Shot=5-shot2023.12 | 91.04 | — | — | — | — | |
| CTXPublished In=NeurIPS 2020, Backbone=ResNet-12, Shot=5-shot2025.12 | 91.01 | — | — | — | — | |
| LR with DCWay=5-way, Shot=5-shot2023.12 | 90.67 | — | — | — | — | |
| Neg-Cosineshots=5-shot, backbone=ResNet-182020.03 | 89.4 | — | — | — | — | |
| NegCosineWay=5-way, Shot=5-shot2023.12 | 89.4 | — | — | — | — | |
| MetaQDAWays=5, Shots=1, Setting=Universal Meta-Learning2023.10 | 88.3 | — | — | — | — | |
| DeepEMDPublished In=CVPR 2020, Backbone=ResNet-12, Shot=5-shot2025.12 | 88.23 | — | — | — | — | |
| MetaOptWays=5, Shots=1, Setting=Universal Meta-Learning, Backbone Status=Frozen2023.10 | 87.9 | — | — | — | — | |
| SNAILWays=5, Shots=1, Setting=Universal Meta-Learning2023.10 | 87.5 | — | — | — | — | |
| ProtoNetshots=5-shot, backbone=ResNet-182020.03 | 87.42 | — | — | — | — | |
| Neg-Softmaxshots=5-shot, backbone=ResNet-182020.03 | 87.3 | — | — | — | — | |
| MetaOptNetBackbone=ResNet12, Shots=5, Ways=52020.10 | 87.09 | — | — | — | — | |
| CausalFSFGPublished In=-, Backbone=ResNet-12, Shot=1-shot2025.12 | 87.05 | — | — | — | — | |
| ProtoNetWays=5, Shots=1, Setting=Universal Meta-Learning, Backbone Status=Frozen2023.10 | 87 | — | — | — | — | |
| ProtoNetWay=5-way, Shot=5-shot2023.12 | 86.64 | — | — | — | — | |
| ProtoNetBackbone=ResNet12, Shots=5, Ways=52020.10 | 86.37 | — | — | — | — | |
| noHubFeature Extractor=ResNet-18, Shot=1-shot, Classifier=SIAMESE2023.03 | 85.88 | — | — | — | — | |
| noHub-SFeature Extractor=ResNet-18, Shot=1-shot, Classifier=SIAMESE2023.03 | 85.81 | — | — | — | — | |
| Delta-encoderBackbone=ResNet12, Shots=5, Ways=52020.10 | 85.6 | — | — | — | — | |
| Bi-FRNPublished In=AAAI 2023, Backbone=ResNet-12, Shot=1-shot2025.12 | 85.44 | — | — | — | — | |
| FRN+TDMPublished In=CVPR 2022, Backbone=ResNet-12, Shot=1-shot2025.12 | 85.15 | — | — | — | — | |
| FRNPublished In=CVPR 2021, Backbone=ResNet-12, Shot=1-shot2025.12 | 84.3 | — | — | — | — | |
| RelationNetBackbone=ResNet12, Shots=5, Ways=52020.10 | 84.28 | — | — | — | — | |
| MatchingNetBackbone=ResNet12, Shots=5, Ways=52020.10 | 84.14 | — | — | — | — | |
| TriNetshot=5-shot, scenario=5-way, semantics=labels, descriptions and attributes2019.06 | 84.1 | — | — | — | — | |
| MatchingNetshots=5-shot, backbone=ResNet-182020.03 | 83.64 | — | — | — | — | |
| Baseline++shots=5-shot, backbone=ResNet-182020.03 | 83.58 | — | — | — | — | |
| Dist. ensembleshot=5-shot, scenario=5-way, semantics=none2019.06 | 83.5 | — | — | — | — | |
| Multiple-Semanticsshot=5-shot, scenario=5-way, semantics=labels, descriptions and attributes2019.06 | 82.9 | — | — | — | — | |
| Baselineshots=5-shot, backbone=ResNet-182020.03 | 82.85 | — | — | — | — | |
| RelationNetshots=5-shot, backbone=ResNet-182020.03 | 82.75 | — | — | — | — | |
| MAMLshots=5-shot, backbone=ResNet-182020.03 | 82.7 | — | — | — | — | |
| Delta-encodershot=5-shot, scenario=5-way, semantics=none2019.06 | 82.6 | — | — | — | — | |
| BaselineBackbone=ResNet12, Shots=5, Ways=52020.10 | 82.54 | — | — | — | — | |
| MTLBackbone=ResNet12, Shots=5, Ways=52020.10 | 82.29 | — | — | — | — | |
| MAMLBackbone=ResNet12, Shots=5, Ways=52020.10 | 82.08 | — | — | — | — | |
| TEENWay=5-way, Shot=1-shot2023.12 | 81.44 | — | — | — | — | |
| SimpleShotshot=5-shot, scenario=5-way, semantics=none, source=re-run of original code2019.06 | 81.3 | — | — | — | — | |
| ProtoNetPublished In=NeurIPS 2017, Backbone=ResNet-12, Shot=1-shot2025.12 | 81.02 | — | — | — | — | |
| Baseline++Backbone=ResNet12, Shots=5, Ways=52020.10 | 81.02 | — | — | — | — | |
| CTXPublished In=NeurIPS 2020, Backbone=ResNet-12, Shot=1-shot2025.12 | 80.39 | — | — | — | — | |
| LR with DCWay=5-way, Shot=1-shot2023.12 | 79.56 | — | — | — | — | |
| Variational Feature DisentanglingBackbone=ResNet12, Shots=1, Ways=52020.10 | 79.12 | — | — | — | — | |
| ProtoNetWays=5, Shots=5, Setting=Universal Meta-Learning, Backbone Status=Unfrozen2023.10 | 77.3 | — | — | — | — | |
| DEML+Meta-SGDshot=5-shot, scenario=5-way, semantics=none2019.06 | 77.1 | — | — | — | — | |
| Multiple-Semanticsshot=1-shot, scenario=5-way, semantics=labels, descriptions and attributes2019.06 | 76.1 | — | — | — | — | |
| ProtoNet+1S-DAug-1Backbone=ViTSmall, Evaluation Protocol=5-way 1-shot2026.01 | 75.72 | — | — | — | — | |
| DeepEMDPublished In=CVPR 2020, Backbone=ResNet-12, Shot=1-shot2025.12 | 75.59 | — | — | — | — | |
| MetaOptNetBackbone=ResNet12, Shots=1, Ways=52020.10 | 75.15 | — | — | — | — | |
| GPICLWays=5, Shots=1, Setting=Universal Meta-Learning2023.10 | 75.1 | — | — | — | — | |
| Delta-encoderBackbone=ResNet12, Shots=1, Ways=52020.10 | 73.91 | — | — | — | — | |
| MTLBackbone=ResNet12, Shots=1, Ways=52020.10 | 73.31 | — | — | — | — | |
| ProtoNetWay=5-way, Shot=1-shot2023.12 | 72.99 | — | — | — | — | |
| Neg-Cosineshots=1-shot, backbone=ResNet-182020.03 | 72.66 | — | — | — | — | |
| NegCosineWay=5-way, Shot=1-shot2023.12 | 72.66 | — | — | — | — | |
| MatchingNetBackbone=ResNet12, Shots=1, Ways=52020.10 | 72.62 | — | — | — | — | |
| MatchingNetshots=1-shot, backbone=ResNet-182020.03 | 72.36 | — | — | — | — | |
| ProtoNetBackbone=ViTSmall, Evaluation Protocol=5-way 1-shot2026.01 | 71.9 | — | — | — | — | |
| ProtoNetshots=1-shot, backbone=ResNet-182020.03 | 71.88 | — | — | — | — | |
| ProtoNet+1S-DAug-1Backbone=SwinTiny, Evaluation Protocol=5-way 1-shot2026.01 | 71.76 | — | — | — | — | |
| ProtoNetBackbone=ResNet12, Shots=1, Ways=52020.10 | 71.57 | — | — | — | — | |
| MetaOptWays=5, Shots=1, Setting=Universal Meta-Learning, Backbone Status=Unfrozen2023.10 | 71.5 | — | — | — | — | |
| Neg-Softmaxshots=1-shot, backbone=ResNet-182020.03 | 71.48 | — | — | — | — | |
| FWT+wave-SANShots=5, Way=52022.03 | 71.16 | — | — | — | — | |
| MAMLBackbone=ResNet12, Shots=1, Ways=52020.10 | 71.11 | — | — | — | — | |
| SimpleShotshot=1-shot, scenario=5-way, semantics=none, source=re-run of original code2019.06 | 70.9 | — | — | — | — | |
| GNN+wave-SANShots=5, Way=52022.03 | 70.31 | — | — | — | — | |
| RelationNetBackbone=ResNet12, Shots=1, Ways=52020.10 | 70.2 | — | — | — | — | |
| MAMLshots=1-shot, backbone=ResNet-182020.03 | 69.96 | — | — | — | — | |
| Delta-encodershot=1-shot, scenario=5-way, semantics=none2019.06 | 69.8 | — | — | — | — | |
| ProtoNetBackbone=SwinTiny, Evaluation Protocol=5-way 1-shot2026.01 | 69.78 | — | — | — | — | |
| TriNetshot=1-shot, scenario=5-way, semantics=labels, descriptions and attributes2019.06 | 69.6 | — | — | — | — | |
| Dist. ensembleshot=1-shot, scenario=5-way, semantics=none2019.06 | 68.7 | — | — | — | — | |
| Baseline++Backbone=ResNet12, Shots=1, Ways=52020.10 | 68.46 | — | — | — | — | |
| RelationNetshots=1-shot, backbone=ResNet-182020.03 | 67.59 | — | — | — | — | |
| Baseline++shots=1-shot, backbone=ResNet-182020.03 | 67.02 | — | — | — | — | |
| FWTShots=5, Way=52022.03 | 66.98 | — | — | — | — | |
| DEML+Meta-SGDshot=1-shot, scenario=5-way, semantics=none2019.06 | 66.9 | — | — | — | — | |
| ATAShots=5, Way=52022.03 | 66.22 | — | — | — | — | |
| Diverse 20 FullShot count=5, Training Dataset=mini-ImageNet, Ensemble Type=Full, Ensemble Size=20, Diversity-encouraged=true2019.03 | 66.17 | — | — | — | — | |
| HyperShotshot=1, ways=5, finetuning=true2022.03 | 66.13 | — | — | — | — | |
| MixStyleShots=5, Way=5, Reproduced=true2022.03 | 65.73 | — | — | — | — | |
| Linear ClassifierShot count=5, Training Dataset=mini-ImageNet2019.03 | 65.57 | — | — | — | — |