Natural Image Classification on ANIMAL10N Real-world noisy labels (test)
92.3AccuracyCUFIT
Evaluation results
| Method | Links | |
|---|---|---|
| CUFITBackbone=DINOv2, Noise rate=0.082024.11 | 92.3 | |
| Co-teachingBackbone=DINOv2, Noise rate=0.082024.11 | 92.2 | |
| JoCorBackbone=DINOv2, Noise rate=0.082024.11 | 91.9 | |
| CoDisBackbone=DINOv2, Noise rate=0.082024.11 | 91.7 | |
| Linear probingBackbone=DINOv2, Noise rate=0.082024.11 | 89.1 | |
| ReinBackbone=DINOv2, Noise rate=0.082024.11 | 88 | |
| Full-trainingBackbone=DINOv2, Noise rate=0.082024.11 | 74.5 |