5-way Few-shot Image Classification on tieredImageNet 5-shot (test)
87.13AccuracynoHub-S
Evaluation results
| Method | Links | ||
|---|---|---|---|
| noHub-SBackbone=ResNet182023.03 | 87.13 | 6.67 | |
| noHubBackbone=WideRes28-102023.03 | 87.07 | 5.5 | |
| noHub-SBackbone=WideRes28-102023.03 | 87.05 | 5.83 | |
| noHubBackbone=ResNet182023.03 | 86.9 | 4.5 | |
| TCPRBackbone=ResNet182023.03 | 86.54 | 3.83 | |
| Robust 20 FullInput size=224, Network=ResNet2019.03 | 86.49 | — | |
| CL2Backbone=ResNet182023.03 | 86.43 | 3 | |
| EASEBackbone=ResNet182023.03 | 86.4 | 3.17 | |
| TCPRBackbone=WideRes28-102023.03 | 86.37 | 4 | |
| L2Backbone=WideRes28-102023.03 | 86.29 | 3.83 | |
| EASEBackbone=WideRes28-102023.03 | 86.29 | 3.33 | |
| L2Backbone=ResNet182023.03 | 86.05 | 1.83 | |
| CL2Backbone=WideRes28-102023.03 | 85.47 | 1.5 | |
| Robust 20-distInput size=224, Network=ResNet2019.03 | 85.43 | — | |
| NoneBackbone=WideRes28-102023.03 | 84.1 | 1.67 | |
| Mean CentroidInput size=224, Network=ResNet2019.03 | 83.89 | — | |
| TADAMInput size=84, Network=ResNet2019.03 | 81.92 | — | |
| MetaOptNet-SVM-trainvalbackbone=ResNet-12, training_set=meta-training + meta-validation2019.04 | 81.75 | — | |
| MetaOptNet-SVMbackbone=ResNet-122019.04 | 81.56 | — | |
| ReRepBackbone=WideRes28-102023.03 | 81.51 | 1.67 | |
| LEObackbone=WRN-28-102019.04 | 81.44 | — | |
| LEOInput size=80, Network=WideResNet2019.03 | 81.44 | — | |
| MetaOptNet-RRbackbone=ResNet-122019.04 | 81.34 | — | |
| ZNBackbone=WideRes28-102023.03 | 75.34 | 5 | |
| ZNBackbone=ResNet182023.03 | 75.32 | 3.83 | |
| ReRepBackbone=ResNet182023.03 | 74.52 | 1.83 | |
| NoneBackbone=ResNet182023.03 | 74.38 | 0.67 | |
| HTChannels=642022.01 | 73.9 | — | |
| Transductive Prop Netsbackbone=64-64-64-642019.04 | 73.3 | — | |
| RFSChannels=642022.01 | 73.2 | — | |
| Prototypical Networksbackbone=64-64-64-642019.04 | 72.69 | — | |
| Relation Networksbackbone=64-96-128-2562019.04 | 71.32 | — | |
| HTChannels=482022.01 | 71.1 | — | |
| MAMLbackbone=32-32-32-322019.04 | 70.3 | — | |
| HTChannels=322022.01 | 70.2 | — | |
| RFSChannels=482022.01 | 69.3 | — | |
| RFSChannels=322022.01 | 66.3 | — | |
| HTChannels=162022.01 | 65.8 | — | |
| RFSChannels=162022.01 | 62 | — | |
| HTChannels=82022.01 | 61.9 | — | |
| RFSChannels=82022.01 | 55.5 | — |