Long-tailed Image Classification on CIFAR-100-LT Imbalanced factor 200 (test)
47.8Top-1 AccuracyPaCo + BatchFormer
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PaCo + BatchFormer2022.03 | 47.8 | — | 68.1 | 53 | 28.2 | |
| Balanced Softmax + BatchFormerTraining stages=one stage2022.03 | 47.5 | — | 70.2 | 53.3 | 25.5 | |
| PaCo2022.03 | 47.1 | — | 68.1 | 51.5 | 27.5 | |
| Balanced SoftmaxTraining stages=one stage2022.03 | 46.4 | — | 70 | 51.5 | 24.3 | |
| SEQLBackbone=ResNet-32, Imbalanced factor=200, gamma=0.95, lambda=3.0 x 10^-32020.03 | 43.38 | 71.94 | — | — | — | |
| Meta-Weight NetBackbone=ResNet-32, Imbalanced factor=2002020.03 | 37.91 | — | — | — | — | |
| Class BalancedBackbone=ResNet-32, Imbalanced factor=2002020.03 | 36.23 | — | — | — | — | |
| Focal LossBackbone=ResNet-32, Imbalanced factor=2002020.03 | 35.62 | — | — | — | — |