Text-to-image retrieval on NoCaps
76.2Recall@1InfoMetIC+
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| InfoMetIC+combined with CLIP similarity=true2023.05 | 76.2 | 95.9 | 98.4 | |
| VladVABackbone=LLAVA-1.5-7B, Zero-shot=true2024.12 | 72.3 | — | 96.5 | |
| InfoMetIC2023.05 | 71.6 | 94.4 | 97.7 | |
| EVA-CLIPParameters=18B, Zero-shot=true2024.12 | 69.3 | — | 94.8 | |
| CLIP-S2023.05 | 67.5 | 91.5 | 95.8 | |
| VLM2VecBackbone=Mistral-7B, Zero-shot=true2024.12 | 65.9 | — | 94.3 | |
| E5-VBackbone=LLAVA-Next-8B, Zero-shot=true2024.12 | 65.9 | — | 94.5 | |
| EVA-CLIPParameters=8B, Zero-shot=true2024.12 | 65.3 | — | 93.2 | |
| OpenCLIPBackbone=ViT-BigG/14, Zero-shot=true2024.12 | 65.1 | — | 93.5 | |
| EVA-02-CLIPBackbone=ViT-E/14+, Zero-shot=true2024.12 | 64.5 | — | 92.9 | |
| OpenCLIPBackbone=ViT-G/14, Zero-shot=true2024.12 | 63.7 | — | 93.2 | |
| BLIP2Backbone=ViT-L, Zero-shot=true2024.12 | 63 | — | 93.8 | |
| BLIPBackbone=ViT-L, Zero-shot=true2024.12 | 62.3 | — | 93.4 | |
| E5-VBackbone=LLAVA-1.5-7B, Zero-shot=true2024.12 | 62 | — | 93 | |
| CLIPBackbone=ViT-L, Zero-shot=true2024.12 | 48.6 | — | 85.7 | |
| LLaVA-1.5-7BZero-shot=true2024.12 | 46.9 | — | 83.3 | |
| TIGER2023.05 | 22.5 | 66.5 | 81.9 |