Few-Shot Classification on mini-ImageNet (test)
92.18AccuracyUnits-NTS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Units-NTSMeta-Training Dataset=mini-ImageNet, Uncertainty Threshold=0.12022.03 | 92.18 | — | — | |
| SemFew-TransBackbone=Swin-T, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 86.49 | — | — | |
| FewTUREBackbone=Swin-T, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 86.38 | — | — | |
| FGFLBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 86.01 | — | — | |
| SP-CLIPBackbone=Visformer-T, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 83.42 | — | — | |
| SVAE-ProtoBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 83.28 | — | — | |
| Units-NTSMeta-Training Dataset=mini-ImageNet, Uncertainty Threshold=0.22022.03 | 83.27 | — | — | |
| SUNBackbone=ViT-S, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 83.25 | — | — | |
| SemFewBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 83.04 | — | — | |
| FEATBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 82.05 | — | — | |
| AM3-BERTBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 81.29 | — | — | |
| ProtoNetBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 80.53 | — | — | |
| CTMBackbone=ResNet-18, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 80.51 | — | — | |
| RFSBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 79.64 | — | — | |
| TRAMLBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 79.54 | — | — | |
| CANBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 79.44 | — | — | |
| Meta-BaselineBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 79.26 | — | — | |
| SemFew-TransBackbone=Swin-T, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 78.94 | — | — | |
| MatchNetBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 78.72 | — | — | |
| AM3Backbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 78.1 | — | — | |
| Meta-AdaMBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 77.92 | — | — | |
| SemFewBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 77.63 | — | — | |
| TADAMBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 76.7 | — | — | |
| SVAE-ProtoBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 74.84 | — | — | |
| KTNBackbone=Conv-128, Way=5-way, Shot=5-shot, Method Category=Semantic-Based2023.11 | 74.16 | — | — | |
| FewTUREBackbone=Swin-T, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 72.4 | — | — | |
| SP-CLIPBackbone=Visformer-T, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 72.31 | — | — | |
| Units-NTSMeta-Training Dataset=mini-ImageNet2022.03 | 71.7 | — | — | |
| Bayesian TAMLMeta-Training Dataset=mini-ImageNet2022.03 | 71.46 | — | — | |
| Meta-SGDMeta-Training Dataset=mini-ImageNet2022.03 | 69.95 | — | — | |
| FGFLBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 69.14 | — | — | |
| fo-Proto-MAMLMeta-Training Dataset=mini-ImageNet2022.03 | 68.96 | — | — | |
| TraNFS-3noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.53 | — | — | |
| Mediannoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.45 | — | — | |
| AM3-BERTBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 68.42 | — | — | |
| RNNPnoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.38 | — | — | |
| Euclideannoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.32 | — | — | |
| TraNFS-2noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.29 | — | — | |
| Vanilla ProtoNetnoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.27 | — | — | |
| Absolutenoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.24 | — | — | |
| Cosinenoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.2 | — | — | |
| Oraclenoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 68.18 | — | — | |
| Baseline++noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 67.91 | — | — | |
| SUNBackbone=ViT-S, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 67.8 | — | — | |
| TRAMLBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 67.1 | — | — | |
| FEATBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 66.78 | — | — | |
| Linear Classifiernoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 66.65 | — | — | |
| MAMLMeta-Training Dataset=mini-ImageNet2022.03 | 66.61 | — | — | |
| Oraclenoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 66.08 | — | — | |
| MatchNetBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 65.64 | — | — | |
| AM3Backbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 65.3 | — | — | |
| TraNFS-3noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 65.08 | — | — | |
| TraNFS-2noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 64.74 | — | — | |
| KTNBackbone=Conv-128, Way=5-way, Shot=1-shot, Method Category=Semantic-Based2023.11 | 64.42 | — | — | |
| CTMBackbone=ResNet-18, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 64.12 | — | — | |
| CANBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 63.85 | — | — | |
| Absolutenoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 63.46 | — | — | |
| Cosinenoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 63.46 | — | — | |
| MAMLnoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 63.25 | — | — | |
| Mediannoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 63.19 | — | — | |
| Meta-BaselineBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 63.17 | — | — | |
| Euclideannoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 63.02 | — | — | |
| Oraclenoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 62.6 | — | — | |
| Vanilla ProtoNetnoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 62.43 | — | — | |
| RNNPnoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 62.43 | — | — | |
| ProtoNetBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 62.39 | — | — | |
| Matching Networksnoise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 62.16 | — | — | |
| RFSBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 62.02 | — | — | |
| Baseline++noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 61.87 | — | — | |
| Meta-AdaMBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 59.89 | — | — | |
| TADAMBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 58.5 | — | — | |
| Linear Classifiernoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 58.41 | — | — | |
| MAMLBackbone=ResNet-12, Way=5-way, Shot=5-shot, Method Category=Visual-Based2023.11 | 58.05 | — | — | |
| Oraclenoise_proportion=60%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 56.89 | — | — | |
| TraNFS-3noise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 56.65 | — | — | |
| Matching Networksnoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 56.21 | — | — | |
| Nearest k=5noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 56.15 | — | — | |
| TraNFS-2noise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 56.14 | — | — | |
| Nearest k=1noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 55.91 | — | — | |
| Nearest k=3noise_proportion=0%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 55.29 | — | — | |
| MAMLnoise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 53.28 | — | — | |
| Cosinenoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 52.42 | — | — | |
| Euclideannoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 52.09 | — | — | |
| Absolutenoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 52.06 | — | — | |
| Baseline++noise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 51.87 | — | — | |
| Mediannoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 51.86 | — | — | |
| RNNPnoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 51.62 | — | — | |
| Vanilla ProtoNetnoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 51.41 | — | — | |
| Nearest k=5noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 50.92 | — | — | |
| MAMLBackbone=ResNet-12, Way=5-way, Shot=1-shot, Method Category=Visual-Based2023.11 | 49.24 | — | — | |
| Nearest k=3noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 48.43 | — | — | |
| Nearest k=1noise_proportion=20%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 47.27 | — | — | |
| Linear Classifiernoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 47.23 | — | — | |
| Matching Networksnoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 46.18 | — | — | |
| TraNFS-3noise_proportion=60%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 42.6 | — | — | |
| MAMLnoise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 42.58 | — | — | |
| TraNFS-2noise_proportion=60%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 42.24 | — | — | |
| Nearest k=5noise_proportion=40%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 42.12 | — | — | |
| Cosinenoise_proportion=60%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 39.9 | — | — | |
| Absolutenoise_proportion=60%, shots=5-shot, ways=5-way, noise_type=symmetric label swap2022.04 | 39.78 | — | — |