Fine-grained Visual Perception on V-Star
79AccuracyVIF
Evaluation Results
| Method | Links | |
|---|---|---|
| VIFVariation=with CoVT2026.04 | 79 | |
| CoVT2026.04 | 77.5 | |
| Qwen2.5-VL-7B2026.04 | 76.4 | |
| VIF2026.04 | 50.8 | |
| SiTebackbone=LLaVA-flickr, variant=ratio2025.12 | 50.79 | |
| SiTebackbone=LLaVA, variant=ratio2025.12 | 50.13 | |
| SiTebackbone=LLaVA, variant=rand2025.12 | 49.95 | |
| LLaVAvariant=flickr2025.12 | 49.32 | |
| LLaVAvariant=Baseline2025.12 | 48.34 | |
| IGVA2026.04 | 48.2 | |
| SiTebackbone=LLaVA-flickr, variant=rand2025.12 | 47.28 | |
| LLaVAAugmentation=ViCrop2026.04 | 46.1 | |
| LLaVA-v1.52026.04 | 45 | |
| SiTebackbone=Qwen2, variant=rand2025.12 | 38.92 | |
| SiTebackbone=Qwen2, variant=ratio2025.12 | 38.09 | |
| Qwen2variant=Baseline2025.12 | 37.83 | |
| SiTebackbone=Qwen2-flickr, variant=rand2025.12 | 36.52 | |
| SiTebackbone=Qwen2-flickr, variant=ratio2025.12 | 36.39 | |
| Qwen2variant=flickr2025.12 | 36.13 | |
| mPLUG-Owl22026.04 | 35.6 |