Few-shot classification on VGG-Flowers
99.06AccuracyWiSE-FT
Evaluation results
| Method | Links | |
|---|---|---|
| WiSE-FTNumber of ways=5, Number of shots=52023.10 | 99.06 | |
| FD-AlignNumber of ways=5, Number of shots=52023.10 | 98.95 | |
| CLIPNumber of ways=5, Number of shots=52023.10 | 98.65 | |
| WiSE-FTNumber of ways=5, Number of shots=12023.10 | 94.16 | |
| FD-AlignNumber of ways=5, Number of shots=12023.10 | 93.5 | |
| CLIPNumber of ways=5, Number of shots=12023.10 | 90.88 | |
| Units-NTS 0.1Meta-Training=Aircraft, QuickDraw, and VGG-Flower, Vacuity Threshold=0.12022.03 | 82.78 | |
| Units-NTS 0.2Meta-Training=Aircraft, QuickDraw, and VGG-Flower, Vacuity Threshold=0.22022.03 | 76.84 | |
| Units-NTSMeta-Training=Aircraft, QuickDraw, and VGG-Flower2022.03 | 70.52 | |
| Bayesian TAMLMeta-Training=Aircraft, QuickDraw, and VGG-Flower2022.03 | 67.72 | |
| fo-Proto-MAMLMeta-Training=Aircraft, QuickDraw, and VGG-Flower2022.03 | 65.24 | |
| MAMLMeta-Training=Aircraft, QuickDraw, and VGG-Flower2022.03 | 60.38 | |
| Meta-SGDMeta-Training=Aircraft, QuickDraw, and VGG-Flower2022.03 | 59.41 | |
| oracleearly-stopping=optimal, algorithm_aggregation=averaged over MAML, ProtoNet, Matching Net2025.12 | 40.55 | |
| Neural Coherenceearly-stopping=Neural Coherence, algorithm_aggregation=averaged over MAML, ProtoNet, Matching Net2025.12 | 40.09 | |
| Baselineearly-stopping=validation-based, algorithm_aggregation=averaged over MAML, ProtoNet, Matching Net2025.12 | 39.59 |