Image Classification on Snapshot Serengeti (test)
98AccuracyClust-PSI-PFL
Evaluation results
| Method | Links | |
|---|---|---|
| Clust-PSI-PFLPartition protocol=Similarity, S=0, K=1002025.12 | 98 | |
| CLMode=Centralized2025.12 | 96 | |
| WildIngLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 90.74 | |
| Linear Probe CLIPLoss Function=Cross-entropy, Backbone=ViT-B/162026.01 | 84.84 | |
| CLIP AdapterLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 84.77 | |
| Long-CLIP AdapterLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 83.97 | |
| BioCLIP AdapterLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 80.47 | |
| WildCLIP-LwFLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 69.53 | |
| WildCLIPLoss Function=Contrastive, Backbone=ViT-B/162026.01 | 68.73 | |
| FedProxPartition protocol=Dirichlet, alpha=0.05, K=1002025.12 | 51 | |
| FedAvgPartition protocol=Dirichlet, alpha=0.05, K=1002025.12 | 47 |