Image Classification on Terra Incognita Trans (test)
84.47AccuracyWildCLIP
Evaluation results
| Method | Links | ||
|---|---|---|---|
| WildCLIPBackbone=ViT-B/162026.01 | 84.47 | — | |
| WildCLIP-LwFBackbone=ViT-B/162026.01 | 82.91 | — | |
| WildIngBackbone=ViT-B/162026.01 | 75.8 | — | |
| Long-CLIP AdapterBackbone=ViT-B/162026.01 | 69.15 | — | |
| CLIP AdapterBackbone=ViT-B/162026.01 | 69.1 | — | |
| Linear Probe CLIPBackbone=ViT-B/162026.01 | 67.23 | — | |
| BioCLIP AdapterBackbone=ViT-B/162026.01 | 63.45 | — | |
| WildCLIP-LwFTraining=Snapshot Serengeti2026.01 | 40.17 | 0.34 | |
| WildIngTraining=Snapshot Serengeti2026.01 | 39.96 | 0.36 | |
| Long-CLIPTraining=ShareGPT4V2026.01 | 37.55 | 0.34 | |
| WildCLIPTraining=Snapshot Serengeti2026.01 | 37.52 | 0.31 | |
| CLIPTraining=OpenAI data2026.01 | 34.67 | 0.32 | |
| Long-CLIP AdapterTraining=Snapshot Serengeti2026.01 | 18.73 | 0.19 | |
| CLIP AdapterTraining=Snapshot Serengeti2026.01 | 16.17 | 0.18 | |
| BioCLIPTraining=TREEOFLIFE-10M2026.01 | 14.53 | 0.15 | |
| BioCLIP AdapterTraining=Snapshot Serengeti2026.01 | 8.59 | 0.08 |