Few-Shot Image Classification on miniImagenet (1-shot, 5-shot)
74.741-Shot AccuracyOurs-Cosine
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours-Cosinebackbone=ViT-S2022.03 | 74.74 | 85.66 | |
| Ours-Classifierbackbone=ViT-S2022.03 | 74.62 | 89.19 | |
| PALbackbone=ResNet-122022.03 | 69.37 | 84.4 | |
| COSOCbackbone=ResNet-122022.03 | 69.28 | 85.16 | |
| CSEIbackbone=ResNet-122022.03 | 68.94 | 85.07 | |
| DCbackbone=WRN-28-102022.03 | 68.57 | 82.88 | |
| CNLbackbone=ResNet-122022.03 | 67.96 | 83.36 | |
| DMFbackbone=ResNet-122022.03 | 67.76 | 82.71 | |
| TPMNbackbone=ResNet-122022.03 | 67.64 | 83.44 | |
| MELRbackbone=ResNet-122022.03 | 67.4 | 83.4 | |
| METAQDAbackbone=WRN2022.03 | 67.38 | 84.27 | |
| IEbackbone=ResNet-122022.03 | 67.28 | 84.78 | |
| MN + MCbackbone=ResNet-122022.03 | 67.14 | 83.82 | |
| BMLbackbone=ResNet-122022.03 | 67.04 | 83.63 | |
| DeepEMDbackbone=ResNet-122022.03 | 65.91 | 82.41 | |
| Baseline-Cosinebackbone=ViT-S2022.03 | 52.92 | 65.04 |