Few-shot Image Classification on CUB (novel)
91.551-Shot AccPT+MAP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PT+MAPSetting=Transductive, Backbone=WRN2020.06 | 91.55 | 93.99 | — | |
| Transfer+SGCSetting=Transductive, Backbone=WRN2020.06 | 88.35 | 92.14 | — | |
| BD-CSPNSetting=Transductive, Backbone=WRN2020.06 | 87.45 | 91.74 | — | |
| S2M2_RSetting=Inductive, Backbone=WRN2020.06 | 80.68 | 90.85 | — | |
| PT+NCMSetting=Inductive, Backbone=WRN2020.06 | 80.57 | 91.15 | — | |
| Matching NetworksSetting=Inductive, Backbone=ResNet182020.06 | 73.49 | 84.45 | — | |
| ProtoNetSetting=Inductive, Backbone=ResNet182020.06 | 72.99 | 86.64 | — | |
| MAMLSetting=Inductive, Backbone=ResNet102020.06 | 70.32 | 80.93 | — | |
| Baseline++Setting=Inductive, Backbone=ResNet102020.06 | 69.55 | 85.17 | — | |
| TSP-HeadBackbone=ResNet, Pre-training dataset=MiniImageNet base set2020.11 | 40.75 | 58.32 | 68.61 | |
| HMSBackbone=ResNet, Pre-training dataset=MiniImageNet base set2020.11 | 40.65 | 57.56 | 67.57 | |
| BaselineBackbone=ResNet, Pre-training dataset=MiniImageNet base set2020.11 | 39.72 | 55.45 | 64.45 | |
| SimCLRBackbone=ResNet, Pre-training dataset=MiniImageNet base set2020.11 | 38.91 | 51.88 | 63.76 | |
| MoCo-v2Backbone=ResNet, Pre-training dataset=MiniImageNet base set2020.11 | 37.52 | 51.04 | 59.48 |