Long-tailed Image Classification on CIFAR100-LT Imbalance Ratio 10
85.7AccuracyCVS+ADRW
Evaluation results
| Method | Links | |
|---|---|---|
| CVS+ADRWBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 85.7 | |
| CVSBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 85.5 | |
| VS+TLA+ADRWBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 85.3 | |
| VSBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 85.1 | |
| CE+ADRWBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 84.3 | |
| LDAM+DRWBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 84.2 | |
| CEBackbone=ViT, Training Strategy=Full Fine-Tuning (FFT)2023.10 | 83.8 | |
| CVS+ADRWBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 83.5 | |
| VS+TLA+ADRWBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 83.1 | |
| CVSBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 83.1 | |
| LDAM+DRWBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 83 | |
| CE+ADRWBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 82.9 | |
| VSBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 82.7 | |
| CEBackbone=ViT, Training Strategy=PEFT (LIFT)2023.10 | 80.5 | |
| Prototype Classifier2023.02 | 69.12 | |
| WD + WD + Maxweight regularization=WD + WD + Max2023.02 | 68.67 | |
| NCM2023.02 | 67.82 | |
| CVS+ADRWBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 67.5 | |
| CVSBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 67.3 | |
| VS+TLA+ADRWBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 67 | |
| VSBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 66.8 | |
| CE+ADRWBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 66.7 | |
| WDweight decay=tuned baseline2023.02 | 66.03 | |
| LDAM+DRWBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 65 | |
| CEBackbone=ResNet, Training Strategy=From scratch, Protocol=(c)2023.10 | 64.6 | |
| PaCOself-supervised pre-training=true, aggressive data augmentation=true2023.02 | 64.2 | |
| DRO-LT2023.02 | 63.41 | |
| cRT2023.02 | 57.93 | |
| LogitAdjust2023.02 | 57.74 | |
| LDAM Loss2023.02 | 56.91 | |
| Focal Loss2023.02 | 55.78 | |
| CE2023.02 | 55.71 |