Fine-grained Detection on MMFakeBench 10000 (test)
37.3F1 ScoreREFORM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| REFORMLanguage Model=Florence2-0.3B, Prompt Method=Ours (Sec.3.2), Zero-shot=true2026.03 | 37.3 | 32.1 | 47 | 56.7 | |
| LLaVA-1.6Language Model=Vicuna-13B, Prompt Method=MMD-Agent, Zero-shot=true2026.03 | 34.5 | 42.7 | 37.5 | 37.4 | |
| BLIP2Language Model=FlanT5-XXL, Prompt Method=MMD-Agent, Zero-shot=true2026.03 | 28.8 | 39 | 30.4 | 32.1 | |
| InstructBLIPLanguage Model=Vicuna-13B, Prompt Method=MMD-Agent, Zero-shot=true2026.03 | 24.5 | 32.1 | 28.8 | 27.3 | |
| VILALanguage Model=LLaMA2-13B, Prompt Method=MMD-Agent, Zero-shot=true2026.03 | 24 | 30.4 | 25.5 | 29.4 | |
| LLaVA-1.6Language Model=Vicuna-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 19 | 16.5 | 26.9 | 32.3 | |
| BLIP2Language Model=FlanT5-XL, Prompt Method=Standard, Zero-shot=true2026.03 | 16.7 | 17.3 | 27.7 | 33.2 | |
| BLIP2Language Model=FlanT5-XXL, Prompt Method=Standard, Zero-shot=true2026.03 | 16.3 | 34.6 | 27.3 | 32.8 | |
| mPLUG-Owl2Language Model=LLaMA2-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 15.1 | 25.2 | 26.3 | 31.5 | |
| LLaVA-1.6Language Model=Vicuna-13B, Prompt Method=Standard, Zero-shot=true2026.03 | 14.4 | 35.7 | 26 | 31.2 | |
| InstructBLIPLanguage Model=Vicuna-13B, Prompt Method=Standard, Zero-shot=true2026.03 | 13.9 | 25.5 | 24.3 | 29.1 | |
| PandaGPTLanguage Model=Vicuna-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 11.6 | 8.6 | 25 | 30 | |
| VILALanguage Model=LLaMA2-13B, Prompt Method=Standard, Zero-shot=true2026.03 | 11.6 | 32.5 | 25 | 30 | |
| VILALanguage Model=LLaMA2-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 11.5 | 7.5 | 25 | 30 | |
| InstructBLIPLanguage Model=Vicuna-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 8.1 | 16.4 | 7.2 | 8.5 | |
| Qwen-VLLanguage Model=Qwen-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 8 | 35.9 | 25.5 | 11.6 | |
| MiniGPT4Language Model=Vicuna-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 5.3 | 6.9 | 21 | 9.1 | |
| Otter-ImageLanguage Model=MPT-7B, Prompt Method=Standard, Zero-shot=true2026.03 | 4.9 | 9.3 | 3.3 | 4 |