Few-shot Image Classification on ISIC CDFSL (test)
48.11Accuracy (5-way 5-shot)Pre+Linear
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Pre+LinearUnSup=false, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 48.11 | 59.31 | 66.48 | |
| Pre+Mean-CentroidUnSup=false, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 47.16 | 56.4 | 61.57 | |
| ProtoTransferUnSup=true, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 45.19 | 59.07 | 66.15 | |
| ProtoNetUnSup=false, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 39.57 | 49.5 | 51.99 | |
| UMTRA-ProtoNetUnSup=true, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 39.21 | 44.62 | 46.48 | |
| UMTRA-ProtoTuneUnSup=true, Backbone=ResNet-10, Training Dataset=mini-ImageNet2020.06 | 38.47 | 51.6 | 60.12 |