Noisy-label detection on ImageNet 100
80F1 ScoreData-OOB
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Data-OOBNoise Rate=30%2024.04 | 80 | — | — | — | |
| Data-OOBNoise Rate=20%2024.04 | 79 | — | — | — | |
| NDDVNoise Rate=30%2024.04 | 78 | — | — | — | |
| NDDVNoise Rate=20%2024.04 | 77 | — | — | — | |
| NDDVNoise Rate=10%2024.04 | 76 | — | — | — | |
| NDDVNoise Rate=5%2024.04 | 74 | — | — | — | |
| Data-OOBNoise Rate=10%2024.04 | 74 | — | — | — | |
| NDDVNoise Rate=40%2024.04 | 74 | — | — | — | |
| Data-OOBNoise Rate=40%2024.04 | 73 | — | — | — | |
| NDDVNoise Rate=45%2024.04 | 67 | — | — | — | |
| Data-OOBNoise Rate=45%2024.04 | 63 | — | — | — | |
| Data-OOBNoise Rate=5%2024.04 | 62 | — | — | — | |
| Beta ShapleyNoise Rate=45%2024.04 | 62 | — | — | — | |
| KNN ShapleyNoise Rate=40%2024.04 | 60 | — | — | — | |
| AMENoise Rate=40%2024.04 | 58 | — | — | — | |
| Beta ShapleyNoise Rate=40%2024.04 | 56 | — | — | — | |
| KNN ShapleyNoise Rate=45%2024.04 | 56 | — | — | — | |
| KNN ShapleyNoise Rate=30%2024.04 | 55 | — | — | — | |
| Data ShapleyNoise Rate=40%2024.04 | 55 | — | — | — | |
| LOONoise Rate=45%2024.04 | 55 | — | — | — | |
| Data ShapleyNoise Rate=45%2024.04 | 55 | — | — | — | |
| LOONoise Rate=40%2024.04 | 54 | — | — | — | |
| Data ShapleyNoise Rate=30%2024.04 | 52 | — | — | — | |
| Beta ShapleyNoise Rate=30%2024.04 | 51 | — | — | — | |
| Data BanzhafNoise Rate=40%2024.04 | 48 | — | — | — | |
| Data BanzhafNoise Rate=45%2024.04 | 48 | — | — | — | |
| Influence FunctionNoise Rate=45%2024.04 | 48 | — | — | — | |
| AMENoise Rate=30%2024.04 | 46 | — | — | — | |
| Influence FunctionNoise Rate=40%2024.04 | 46 | — | — | — | |
| KNN ShapleyNoise Rate=20%2024.04 | 45 | — | — | — | |
| Data BanzhafNoise Rate=30%2024.04 | 42 | — | — | — | |
| Influence FunctionNoise Rate=30%2024.04 | 42 | — | — | — | |
| LOONoise Rate=30%2024.04 | 39 | — | — | — | |
| Data BanzhafNoise Rate=20%2024.04 | 31 | — | — | — | |
| Influence FunctionNoise Rate=20%2024.04 | 31 | — | — | — | |
| KNN ShapleyNoise Rate=10%2024.04 | 30 | — | — | — | |
| LOONoise Rate=20%2024.04 | 30 | — | — | — | |
| AMENoise Rate=45%2024.04 | 27 | — | — | — | |
| Data ShapleyNoise Rate=20%2024.04 | 25 | — | — | — | |
| Beta ShapleyNoise Rate=20%2024.04 | 25 | — | — | — | |
| Data ShapleyNoise Rate=10%2024.04 | 19 | — | — | — | |
| Beta ShapleyNoise Rate=10%2024.04 | 19 | — | — | — | |
| Data BanzhafNoise Rate=10%2024.04 | 18 | — | — | — | |
| Influence FunctionNoise Rate=10%2024.04 | 18 | — | — | — | |
| AMENoise Rate=10%2024.04 | 18 | — | — | — | |
| KNN ShapleyNoise Rate=5%2024.04 | 17 | — | — | — | |
| LOONoise Rate=10%2024.04 | 16 | — | — | — | |
| Data ShapleyNoise Rate=5%2024.04 | 12 | — | — | — | |
| Beta ShapleyNoise Rate=5%2024.04 | 11 | — | — | — | |
| Influence FunctionNoise Rate=5%2024.04 | 11 | — | — | — | |
| LOONoise Rate=5%2024.04 | 9 | — | — | — | |
| Data BanzhafNoise Rate=5%2024.04 | 9 | — | — | — | |
| AMENoise Rate=5%2024.04 | 1 | — | — | — | |
| AMENoise Rate=20%2024.04 | 1 | — | — | — | |
| 2D geometric metricBackbone=Resnet50, Label error(%)=5%2026.06 | — | 80.47 | 20.31 | 99.37 | |
| 2D geometric metricBackbone=Resnet50, Label error(%)=10%2026.06 | — | 81.67 | 35.29 | 99.87 | |
| 2D geometric metricBackbone=Resnet50, Label error(%)=20%2026.06 | — | 88.35 | 63.25 | 99.71 | |
| 2D geometric metricBackbone=Resnet50, Label error(%)=40%2026.06 | — | 97.5 | 95.75 | 98.11 | |
| 2D geometric metricBackbone=Vit, Label error(%)=5%2026.06 | — | 77.6 | 18.06 | 98.4 | |
| 2D geometric metricBackbone=Vit, Label error(%)=10%2026.06 | — | 77.79 | 31.01 | 99.66 | |
| 2D geometric metricBackbone=Vit, Label error(%)=20%2026.06 | — | 90.01 | 67.14 | 68.03 | |
| 2D geometric metricBackbone=Vit, Label error(%)=40%2026.06 | — | 96.03 | 92.64 | 97.86 | |
| 3D multi-metricBackbone=Resnet50, Label error(%)=5%2026.06 | — | 80.21 | 20.12 | 99.62 | |
| 3D multi-metricBackbone=Resnet50, Label error(%)=10%2026.06 | — | 83.81 | 38.17 | 99.83 | |
| 3D multi-metricBackbone=Resnet50, Label error(%)=20%2026.06 | — | 97.98 | 91.94 | 98.55 | |
| 3D multi-metricBackbone=Resnet50, Label error(%)=40%2026.06 | — | 97.69 | 95.7 | 98.65 | |
| 3D multi-metricBackbone=Vit, Label error(%)=5%2026.06 | — | 75.72 | 17.02 | 99.46 | |
| 3D multi-metricBackbone=Vit, Label error(%)=10%2026.06 | — | 79.02 | 32.25 | 99.75 | |
| 3D multi-metricBackbone=Vit, Label error(%)=20%2026.06 | — | 92.71 | 74.15 | 97.56 | |
| 3D multi-metricBackbone=Vit, Label error(%)=40%2026.06 | — | 96.18 | 92.62 | 98.29 | |
| CleanlabLabel error (%)=5%2026.06 | — | 86.92 | 25.41 | 83.43 | |
| CleanlabLabel error (%)=10%2026.06 | — | 85.6 | 39.7 | 84.82 | |
| CleanlabLabel error (%)=20%2026.06 | — | 84.45 | 57.41 | 86.19 | |
| CleanlabLabel error (%)=40%2026.06 | — | 81.32 | 72.87 | 84.9 |