Image Classification on ImageNet New classes
73.92New Accuracy (%)TEn
Evaluation Results
| Method | Links | |
|---|---|---|
| TEnmodels=(BLIP, BLIP2, CLIP-ViT-B/32)2023.11 | 73.92 | |
| TFEnmodels=(BLIP, BLIP2, CLIP-ViT-B/32)2023.11 | 71.5 | |
| FARLProtocol=Base-to-novel generalization2025.12 | 71.33 | |
| MMAProtocol=Base-to-novel generalization2025.12 | 71.23 | |
| MMRLProtocol=Base-to-novel generalization2025.12 | 71.13 | |
| MetaPromptProtocol=Base-to-novel generalization2025.12 | 70.83 | |
| MaPLeProtocol=Base-to-novel generalization2025.12 | 70.54 | |
| CoCoOpProtocol=Base-to-novel generalization2025.12 | 70.43 | |
| Mean Ensemblemodels=(BLIP, BLIP2, CLIP-ViT-B/32)2023.11 | 70 | |
| KgCoOpProtocol=Base-to-novel generalization2025.12 | 69.96 | |
| PLOTProtocol=Base-to-novel generalization2025.12 | 69.87 | |
| TCPProtocol=Base-to-novel generalization2025.12 | 69.87 | |
| ProVPProtocol=Base-to-novel generalization2025.12 | 69.21 | |
| CLIPProtocol=Base-to-novel generalization2025.12 | 68.14 | |
| TEnmodels=(BLIP, BLIP2)2023.11 | 66.83 | |
| CoOpProtocol=Base-to-novel generalization2025.12 | 66.31 | |
| CLIP-ViT-B/32mode=Zero-shot2023.11 | 64.02 | |
| TFEnmodels=(BLIP, BLIP2)2023.11 | 63.23 | |
| Mean Ensemblemodels=(BLIP, BLIP2)2023.11 | 61.36 | |
| BLIPmode=Zero-shot2023.11 | 55.75 | |
| BLIP2mode=Zero-shot2023.11 | 51.98 |