Few-Shot Class-Incremental Learning on ImageNet-R (10-way 5-shot, test)
84.6Accuracy (Session 0)TEEN
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TEENBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 84.6 | 76.7 | 68.8 | 67.6 | 64.3 | 60.6 | 58.3 | 56.1 | 56.1 | 54.7 | 54.9 | 63.9 | |
| CASPBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 83.7 | 80.8 | 80.5 | 78.2 | 77.4 | 76.5 | 75.4 | 75.2 | 75.1 | 74.5 | 73.8 | 77.4 | |
| ASPBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 83.3 | 80.4 | 79.6 | 77 | 75.6 | 74.7 | 73 | 72.1 | 71.9 | 70.9 | 69.7 | 75.3 | |
| SEC-FBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 82.9 | 80.5 | 79.8 | 77.5 | 76.7 | 76.3 | 74.8 | 74.6 | 74.1 | 73.8 | 73 | 76.7 | |
| Coda-PBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 82.8 | 80.1 | 78.6 | 75.3 | 73.1 | 71.8 | 70 | 69.2 | 68.8 | 67.4 | 66.4 | 73 | |
| PriViLegeBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 80 | 77.3 | 77 | 73.8 | 72.8 | 72.3 | 71.3 | 69.7 | 69.5 | 67.7 | 66.5 | 72.5 | |
| L2PBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 73.9 | 70.9 | 69.3 | 65.9 | 64 | 62.6 | 60.1 | 59.5 | 59 | 58.2 | 56.8 | 63.7 | |
| DualPBackbone=ViT-B/16-1K, Setup=10-way 5-shot2026.01 | 71.5 | 69 | 69 | 67.4 | 66.6 | 65.9 | 64.1 | 64 | 64 | 63.4 | 62.5 | 66.1 |