Text-to-Image Retrieval on CC3M
45.7RecallCLIP-RD
Evaluation results
| Method | Links | |
|---|---|---|
| CLIP-RDBackbone=ResNet-50, Zero-shot=true2026.03 | 45.7 | |
| CLIP-KDBackbone=ResNet-50, Zero-shot=true2026.03 | 45.3 | |
| ViT-B/16Backbone=ViT-B/16, Zero-shot=true, Role=Teacher2026.03 | 42.2 | |
| CLIP-RDBackbone=EfficientNet-B0, Zero-shot=true2026.03 | 41 | |
| ResNet-50Backbone=ResNet-50, Zero-shot=true, Role=Student2026.03 | 39.6 | |
| EfficientNet-B0Backbone=EfficientNet-B0, Zero-shot=true, Role=Student2026.03 | 36.7 | |
| CLIP-RDBackbone=ResNet-18, Zero-shot=true2026.03 | 34.5 | |
| CLIP-KDBackbone=ResNet-18, Zero-shot=true2026.03 | 34.1 | |
| ResNet-18Backbone=ResNet-18, Zero-shot=true, Role=Student2026.03 | 30.9 |