Few-shot Image Classification (1-shot, 5-way) on FC100 (test)
51.35Average AccuracyR2-D2 (best)
Evaluation results
| Method | Links | |
|---|---|---|
| R2-D2 (best)Task Augmentation=Rotating, Configuration=best (ensemble + val)2020.02 | 51.35 | |
| M-SVM (best)Task Augmentation=Rotating, Configuration=best (ensemble + val)2020.02 | 49.77 | |
| HCTransformers-Cosinebackbone=ViT-S2022.03 | 48.27 | |
| HCTransformers-Classifierbackbone=ViT-S2022.03 | 48.15 | |
| IEbackbone=ResNet-122022.03 | 47.76 | |
| M-SVM2020.02 | 47.2 | |
| PALbackbone=ResNet-122022.03 | 47.2 | |
| TPMNbackbone=ResNet-122022.03 | 46.93 | |
| DeepEMDbackbone=ResNet-122022.03 | 46.47 | |
| MN + MCbackbone=ResNet-122022.03 | 46.4 | |
| MTL2020.02 | 45.1 | |
| BMLbackbone=ResNet-122022.03 | 45 | |
| ALFA+METALbackbone=ResNet-122022.03 | 44.54 | |
| ConstellationNetbackbone=ResNet-122022.03 | 43.8 | |
| MixtFSLbackbone=ResNet-122022.03 | 41.5 | |
| Baseline-Cosinebackbone=ViT-S2022.03 | 40.83 | |
| SimSiam + AugSelfBackbone=ResNet-18, Pre-training Dataset=STL102021.11 | 40.68 | |
| TADAM2020.02 | 40.1 | |
| MoCo v2 + AugSelfBackbone=ResNet-18, Pre-training Dataset=STL102021.11 | 39.66 | |
| SimSiam + AugSelfBackbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 39.37 | |
| ProtoNets2020.02 | 37.5 | |
| SimSiamBackbone=ResNet-18, Pre-training Dataset=STL102021.11 | 36.72 | |
| SimSiamBackbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 36.19 | |
| MoCo v2Backbone=ResNet-18, Pre-training Dataset=STL102021.11 | 35.69 | |
| MoCo v2 + AugSelfBackbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 35.02 | |
| Supervised + AugSelfBackbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 34.7 | |
| SupervisedBackbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 33.15 | |
| MoCo v2Backbone=ResNet-50, Pre-training Dataset=ImageNet1002021.11 | 31.67 |