Skin Lesion Classification on ISIC 3-way 5-shot 2018 (test)
81.16AccuracyHybrid distil
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Hybrid distilBackbone=ResNet-122021.10 | 81.16 | 77.35 | |
| MoCo_v1 (baseline)Backbone=ResNet-122021.10 | 79.01 | 74.47 | |
| MoCo_v2Backbone=ResNet-122021.10 | 75.8 | 70.69 | |
| SRL-distilBackbone=ResNet-122021.10 | 72.78 | 65.89 | |
| SRL-simpleBackbone=ResNet-122021.10 | 70.53 | 65.58 | |
| Hybrid distilBackbone=4 conv blocks2021.10 | 70.26 | 65.89 | |
| SimCLRBackbone=ResNet-122021.10 | 70.18 | 63.73 | |
| MoCo_v1 (baseline)Backbone=4 conv blocks2021.10 | 69.94 | 64.31 | |
| SimCLRBackbone=4 conv blocks2021.10 | 66.25 | 61.75 | |
| MoCo_v2Backbone=4 conv blocks2021.10 | 62.91 | 57.43 | |
| SRL-distilBackbone=4 conv blocks2021.10 | 62.07 | 58.71 | |
| SRL-simpleBackbone=4 conv blocks2021.10 | 60.62 | 53.98 | |
| DAMLBackbone=ResNet-122021.10 | 59.44 | 54.66 | |
| MAMLBackbone=ResNet-122021.10 | 58.94 | 53.51 | |
| UMTRABackbone=ResNet-122021.10 | 57.33 | 53.06 | |
| DAMLBackbone=4 conv blocks2021.10 | 56.11 | 53.29 | |
| MAMLBackbone=4 conv blocks2021.10 | 55.73 | 49.51 | |
| RelationNetBackbone=ResNet-122021.10 | 55.71 | 49.34 | |
| UMTRABackbone=4 conv blocks2021.10 | 54.16 | 49.41 | |
| RelationNetBackbone=4 conv blocks2021.10 | 49.98 | 42.86 | |
| CACTUS-MAMLBackbone=ResNet-122021.10 | 48.11 | 44.32 | |
| CACTUS-ProtoNetsBackbone=ResNet-122021.10 | 47.95 | 44.08 | |
| Training from scratchBackbone=ResNet-122021.10 | 45.36 | 38.41 | |
| CACTUS-MAMLBackbone=4 conv blocks2021.10 | 45.31 | 39.96 | |
| Training from scratchBackbone=4 conv blocks2021.10 | 43.13 | 37.57 | |
| ProtoNetsBackbone=ResNet-122021.10 | 42.45 | 34.92 | |
| ProtoNetsBackbone=4 conv blocks2021.10 | 37.95 | 34.08 | |
| CACTUS-ProtoNetsBackbone=4 conv blocks2021.10 | 36.86 | 32.76 |