Hierarchical Classification on CIFAR-100 (test)
0.7787AccuracyFlamingo-I
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Flamingo-IBackbone=ResNet-502025.03 | 0.7787 | 2.31 | 0.51 | 2.06 | 3.08 | 0.57 | 0.09 | 0.21 | — | |
| Hier-COSBackbone=ResNet-502025.03 | 0.7779 | 2.21 | 0.49 | 1.09 | 2.17 | 0.93 | 0.76 | 0.85 | — | |
| Cross EntropyBackbone=ResNet-502025.03 | 0.7777 | 2.33 | 0.52 | 2.25 | 3.19 | 0.54 | 0.05 | 0.17 | — | |
| HXE+CRMBackbone=ResNet-502025.03 | 0.7763 | 2.3 | 0.51 | 1.11 | 2.18 | — | — | — | — | |
| HAFrameBackbone=ResNet-502025.03 | 0.7753 | 2.24 | 0.5 | 1.12 | 2.18 | 0.92 | 0.72 | 0.83 | — | |
| HAFeat-IBackbone=ResNet-502025.03 | 0.7751 | 2.24 | 0.5 | 1.43 | 2.64 | 0.69 | 0.41 | 0.47 | — | |
| HAFeat-IIBackbone=ResNet-502025.03 | 0.7647 | 2.27 | 0.54 | 1.7 | 2.84 | 0.63 | 0.22 | 0.32 | — | |
| Flamingo-IIBackbone=ResNet-502025.03 | 0.7526 | 2.32 | 0.57 | 2.41 | 3.29 | 0.51 | — | 0.14 | — | |
| Soft-labelsBackbone=ResNet-50, Beta=302025.03 | 0.7301 | 2.38 | 0.64 | 1.39 | 2.79 | — | — | — | — | |
| Barz & DenzlerBackbone=ResNet-502025.03 | 0.6831 | 2.36 | 0.75 | 1.25 | 2.49 | — | — | — | — | |
| YOLO-v2Backbone=ResNet-502025.03 | 0.6797 | 3.72 | 1.19 | 2.85 | 3.39 | — | — | — | — | |
| Soft-labelsBackbone=ResNet-50, Beta=42025.03 | 0.6785 | 2.21 | 0.71 | 1.23 | 2.23 | — | — | — | — |