Image Classification on Caltech-101 (Top-1 and Top-5 Accuracy)
46.92Top-1 AccuracyCE+SIL+SupCon2
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CE+SIL+SupCon2Loss function=Cross-Entropy + Silhouette + SupCon22026.03 | 46.92 | 66.31 | |
| CE+SILLoss function=Cross-Entropy + Silhouette2026.03 | 45.95 | 63.83 | |
| SupCon2Loss function=Supervised Contrastive (two-view)2026.03 | 44.17 | 64.04 | |
| ProxyNCALoss function=Proxy Neighborhood Component Analysis2026.03 | 44.16 | — | |
| CELoss function=Cross-Entropy2026.03 | 44.04 | 63.47 | |
| SupConLoss function=Supervised Contrastive2026.03 | 38.57 | 60.07 | |
| CenterLoss function=Center Loss2026.03 | 38.32 | 57.09 |