Object Hallucination Evaluation on MSCOCO POPE (random popular adversarial)
86.79AccuracySVC
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SVCBase MLLM=LLaVA-1.52026.03 | 86.79 | 87.04 | |
| VISTABase MLLM=LLaVA-1.52026.03 | 86.15 | 86.29 | |
| ONLYBase MLLM=LLaVA-1.52026.03 | 86.03 | 86.22 | |
| PAIBase MLLM=LLaVA-1.52026.03 | 85.85 | 86.08 | |
| ONLYBase MLLM=InstructBLIP2026.03 | 84.9 | 85.01 | |
| VISTABase MLLM=InstructBLIP2026.03 | 84.87 | 84.95 | |
| VCDBase MLLM=InstructBLIP2026.03 | 84.81 | 85.28 | |
| VCDBase MLLM=LLaVA-1.52026.03 | 84.8 | 85.65 | |
| VanillaBase MLLM=LLaVA-1.52026.03 | 84.79 | 85.61 | |
| SVCBase MLLM=InstructBLIP2026.03 | 84.59 | 84.97 | |
| VanillaBase MLLM=InstructBLIP2026.03 | 84.36 | 84.64 | |
| PAIBase MLLM=InstructBLIP2026.03 | 83.98 | 84.02 | |
| SVCBase MLLM=Shikra2026.03 | 83.84 | 83.58 | |
| ONLYBase MLLM=Shikra2026.03 | 82.75 | 82.85 | |
| VISTABase MLLM=Shikra2026.03 | 82.44 | 82.47 | |
| VCDBase MLLM=Shikra2026.03 | 81.34 | 82.25 | |
| VanillaBase MLLM=Shikra2026.03 | 81.32 | 82.01 | |
| PAIBase MLLM=Shikra2026.03 | 81.3 | 80.81 | |
| ONLYBase MLLM=MiniGPT-42026.03 | 76.98 | 77.62 | |
| SVCBase MLLM=MiniGPT-42026.03 | 76.78 | 77.29 | |
| VanillaBase MLLM=MiniGPT-42026.03 | 76.76 | 76.82 | |
| VISTABase MLLM=MiniGPT-42026.03 | 76.06 | 76.8 | |
| VCDBase MLLM=MiniGPT-42026.03 | 76.01 | 76.28 | |
| PAIBase MLLM=MiniGPT-42026.03 | 75.64 | 77.57 |