Image Classification on ImageNet-LT Uniform Shifted (test)
53.1Top-1 AccDiVE
Evaluation results
| Method | Links | |
|---|---|---|
| DiVETraining epochs=90, Post-compensation=false2021.03 | 53.1 | |
| DiVETraining epochs=90, Post-compensation=true2021.03 | 53.1 | |
| LADEPost-prior Correction (PC)=implicit, Training Epochs=1802020.12 | 53 | |
| PC SoftmaxPost-prior Correction (PC)=true, Training Epochs=1802020.12 | 52.8 | |
| Balanced SoftmaxPost-prior Correction (PC)=false, Training Epochs=1802020.12 | 52.1 | |
| PC Balanced SoftmaxPost-prior Correction (PC)=true, Training Epochs=1802020.12 | 52.1 | |
| Causal NormPost-prior Correction (PC)=false, Training Epochs=1802020.12 | 52 | |
| PC Causal NormPost-prior Correction (PC)=true, Training Epochs=1802020.12 | 52 | |
| BSCETraining epochs=90, Post-compensation=false2021.03 | 50.48 | |
| BSCETraining epochs=90, Post-compensation=true2021.03 | 50.48 | |
| CETraining epochs=90, Post-compensation=true2021.03 | 49.33 | |
| SoftmaxPost-prior Correction (PC)=false, Training Epochs=1802020.12 | 48.2 | |
| CETraining epochs=90, Post-compensation=false2021.03 | 43.89 |