Binary image-text classification on KID2
0.86AUROCCapCLIP-B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CapCLIP-BBackbone=ViT-B/16, Setting=Zero-shot2026.05 | 0.86 | 0.842 | |
| CapCLIP-BBackbone=ViT-L/14, Setting=Zero-shot2026.05 | 0.852 | 0.82 | |
| CapCLIP-MIXBackbone=ViT-B/16, Setting=Zero-shot2026.05 | 0.833 | 0.806 | |
| CapCLIP-MBackbone=ViT-L/14, Setting=Zero-shot2026.05 | 0.824 | 0.724 | |
| CapCLIP-MBackbone=ViT-B/16, Setting=Zero-shot2026.05 | 0.8 | 0.75 | |
| CapCLIP-MIXBackbone=ViT-L/14, Setting=Zero-shot2026.05 | 0.689 | 0.615 | |
| CLIP-OpenAISetting=Zero-shot2026.05 | 0.618 | 0.545 | |
| MedImageInsightSetting=Zero-shot2026.05 | 0.612 | 0.545 | |
| BioMedCLIPSetting=Zero-shot2026.05 | 0.57 | 0.508 | |
| MedCLIPSetting=Zero-shot2026.05 | 0.507 | 0.448 |