Visual Question Answering on A-OKVQA POPE Evaluation (Adversarial)
82AccuracyMESA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MESAModel=Qwen-VL2026.04 | 82 | 79.67 | 85.93 | 82.68 | |
| ICTModel=Qwen-VL2026.04 | 81.56 | 79.19 | 84.11 | 81.59 | |
| VCDModel=Qwen-VL2026.04 | 81.27 | 77.79 | 87.53 | 82.38 | |
| VAFModel=Qwen-VL2026.04 | 81.06 | 85.78 | 78.73 | 82.1 | |
| VTIModel=Qwen-VL2026.04 | 80.57 | 79.64 | 82.13 | 80.87 | |
| VanillaModel=Qwen-VL2026.04 | 79.56 | 77.66 | 83 | 80.24 | |
| NullModel=Qwen-VL2026.04 | 79.3 | 77.31 | 82.93 | 80.02 | |
| ICDModel=Qwen-VL2026.04 | 79.17 | 86.61 | 69 | 76.81 | |
| ICTModel=LLaVA-v1.52026.04 | 78.53 | 75.47 | 84.53 | 79.74 | |
| MESAModel=LLaVA-v1.52026.04 | 78.03 | 74.29 | 85.73 | 79.6 | |
| VAFModel=LLaVA-v1.52026.04 | 76.93 | 73.21 | 84.93 | 78.64 | |
| VTIModel=LLaVA-v1.52026.04 | 76.77 | 73.11 | 84.67 | 78.47 | |
| ICDModel=LLaVA-v1.52026.04 | 75.7 | 73.43 | 80.53 | 76.82 | |
| VCDModel=LLaVA-v1.52026.04 | 74.97 | 70.01 | 87.36 | 77.73 | |
| NullModel=LLaVA-v1.52026.04 | 72.13 | 66.58 | 88.86 | 76.12 | |
| VanillaModel=LLaVA-v1.52026.04 | 67.5 | 61.62 | 92.8 | 74.06 |