Object Hallucination on A-OKVQA POPE (test)
90.13Accuracy (Random)OSGA
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OSGAModel Architecture=LLaVA-v1.52026.01 | 90.13 | 89.75 | 85.27 | 85.43 | 79.9 | 78.9 | 85.1 | 84.69 | — | — | — | — | — | — | |
| MAPModel=LLaVA1.5, Decoding Strategy=MAP2025.08 | 89.03 | 89.7 | 82.7 | 84.7 | 72.77 | 77.84 | — | — | — | — | — | — | — | — | |
| VanillaModel=InstBLIP, Decoding Strategy=Vanilla2025.08 | 88.6 | 88.34 | 79.47 | 81.85 | 71.37 | 76.38 | — | — | — | — | — | — | — | — | |
| MAPModel=InstBLIP, Decoding Strategy=MAP2025.08 | 88.6 | 88.41 | 80.2 | 81.85 | 73.03 | 76.8 | — | — | — | — | — | — | — | — | |
| SPINModel=InstBLIP, Decoding Strategy=SPIN2025.08 | 88.57 | 88.11 | 79.6 | 81.97 | 71.47 | 76.47 | — | — | — | — | — | — | — | — | |
| HGAIModel=InstBLIP, Decoding Strategy=HGAI2025.08 | 88.57 | 88.35 | 79.3 | 81.77 | 71.33 | 76.4 | — | — | — | — | — | — | — | — | |
| HGAIModel=mPLUG2, Decoding Strategy=HGAI2025.08 | 88.33 | 88.26 | 84.57 | 85.04 | 77.57 | 79.64 | — | — | — | — | — | — | — | — | |
| MAPModel=mPLUG2, Decoding Strategy=MAP2025.08 | 88.33 | 88.32 | 84.77 | 85.27 | 77.7 | 79.82 | — | — | — | — | — | — | — | — | |
| OPERAModel Architecture=LLaVA-v1.52026.01 | 88.27 | 87.54 | 85.17 | 84.74 | 79.37 | 79.97 | 84.27 | 84.08 | — | — | — | — | — | — | |
| VanillaModel=mPLUG2, Decoding Strategy=Vanilla2025.08 | 88.27 | 88.23 | 84.73 | 85.21 | 77.57 | 79.67 | — | — | — | — | — | — | — | — | |
| DAMOModel=mPLUG2, Decoding Strategy=DAMO2025.08 | 88.13 | 88.1 | 84.43 | 84.95 | 77.23 | 79.42 | — | — | — | — | — | — | — | — | |
| SPINModel=mPLUG2, Decoding Strategy=SPIN2025.08 | 88.13 | 87.96 | 84.43 | 84.77 | 77.6 | 79.46 | — | — | — | — | — | — | — | — | |
| DCLAModel=InstBLIP, Decoding Strategy=DCLA2025.08 | 88.13 | 88.28 | 79.83 | 81.59 | 72.37 | 76.39 | — | — | — | — | — | — | — | — | |
| DCLAModel=LLaVA1.5, Decoding Strategy=DCLA2025.08 | 87.93 | 88.98 | 81.13 | 83.78 | 70.23 | 76.6 | — | — | — | — | — | — | — | — | |
| DAMOModel=LLaVA1.5, Decoding Strategy=DAMO2025.08 | 87.87 | 88.91 | 81.3 | 83.87 | 70.53 | 76.75 | — | — | — | — | — | — | — | — | |
| DCLAModel=mPLUG2, Decoding Strategy=DCLA2025.08 | 87.87 | 88.23 | 83.77 | 84.82 | 75.33 | 78.63 | — | — | — | — | — | — | — | — | |
| SPINModel=LLaVA1.5, Decoding Strategy=SPIN2025.08 | 87.7 | 88.79 | 80.63 | 83.41 | 69.97 | 76.43 | — | — | — | — | — | — | — | — | |
| HGAIModel=LLaVA1.5, Decoding Strategy=HGAI2025.08 | 87.37 | 88.5 | 80.37 | 83.2 | 70 | 76.42 | — | — | — | — | — | — | — | — | |
| DAMOModel=InstBLIP, Decoding Strategy=DAMO2025.08 | 87.33 | 87.77 | 79.07 | 81.29 | 71.4 | 76.07 | — | — | — | — | — | — | — | — | |
| VanillaModel=LLaVA1.5, Decoding Strategy=Vanilla2025.08 | 87.1 | 88.3 | 80.1 | 83.08 | 69.03 | 75.9 | — | — | — | — | — | — | — | — | |
| VCDModel=LLaVA1.5, Decoding Strategy=VCD2025.08 | 86.97 | 87.23 | 78.9 | 81.82 | 67.9 | 75.01 | — | — | — | — | — | — | — | — | |
| VCDModel=mPLUG2, Decoding Strategy=VCD2025.08 | 86.83 | 86.93 | 82.07 | 82.96 | 75.5 | 78.13 | — | — | — | — | — | — | — | — | |
| SDCDModel=LLaVA1.5, Decoding=SDCD2026.01 | 86.3 | 86.03 | 80.93 | 81.56 | 73.93 | 76.14 | — | — | 87.79 | 84.33 | 78.96 | 84.33 | 70.19 | 83.2 | |
| VCDModel Architecture=LLaVA-v1.52026.01 | 86.15 | 86.34 | 81.85 | 82.82 | 74.97 | 77.73 | 80.99 | 82.3 | — | — | — | — | — | — | |
| ICDModel Architecture=LLaVA-v1.52026.01 | 85.57 | 85.06 | 81.93 | 81.95 | 77.43 | 78.99 | 81.64 | 82 | — | — | — | — | — | — | |
| VCDModel=LLaVA1.5, Decoding=VCD2026.01 | 85.03 | 84.61 | 80.6 | 80.92 | 74.9 | 76.59 | — | — | 87.09 | 82.27 | 79.61 | 82.27 | 71.75 | 82.13 | |
| RegularModel=LLaVA1.5, Decoding=Regular2026.01 | 84.03 | 83.22 | 80.23 | 80.03 | 74.27 | 75.34 | — | — | 87.68 | 79.2 | 80.87 | 79.2 | 72.33 | 78.6 | |
| BaselineModel Architecture=LLaVA-v1.52026.01 | 83.45 | 82.56 | 79.9 | 79.59 | 74.04 | 75.15 | 79.13 | 79.1 | — | — | — | — | — | — | |
| VCDModel=InstBLIP, Decoding Strategy=VCD2025.08 | 82.03 | 82.36 | 76.9 | 78.4 | 71.67 | 74.93 | — | — | — | — | — | — | — | — |