Incremental Learning on CUB 200 2011
81.07Accuracy (Session 1)F2M
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| F2MSetup=10-way 5-shot, re-implemented=false2021.10 | 81.07 | 78.16 | 75.57 | 72.89 | 70.86 | 68.17 | 67.01 | 65.26 | 63.36 | 61.76 | 60.26 | — | |
| RebalanceSetup=10-way 5-shot, re-implemented=true2021.10 | 80.94 | 70.32 | 62.96 | 57.19 | 51.06 | 46.7 | 44.03 | 40.15 | 36.75 | 34.88 | 32.09 | — | |
| BaselineSetup=10-way 5-shot, re-implemented=false2021.10 | 80.87 | 77.15 | 74.46 | 72.26 | 69.47 | 67.18 | 65.62 | 63.68 | 61.3 | 59.72 | 58.12 | — | |
| FSLLSetup=10-way 5-shot, re-implemented=true2021.10 | 80.83 | 77.38 | 72.37 | 71.84 | 67.51 | 65.3 | 63.75 | 61.16 | 59.05 | 58.03 | 55.82 | — | |
| iCaRLSetup=10-way 5-shot, re-implemented=true2021.10 | 79.58 | 67.63 | 64.17 | 61.8 | 58.1 | 55.51 | 53.34 | 50.89 | 48.62 | 47.34 | 45.6 | — | |
| Joint full fine-tuningstrategy=full fine-tuning, backbone=ViT2023.07 | — | — | — | — | — | — | — | — | — | — | — | 90.5 | |
| RanPACalgorithm=Algorithm 12023.07 | — | — | — | — | — | — | — | — | — | — | — | 90.3 | |
| SLCAnotes=smaller number of training samples than [65]2023.07 | — | — | — | — | — | — | — | — | — | — | — | 84.7 |