Object Hallucination on MSCOCO POPE (test)
90.7Accuracy (Random)HGAI
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HGAIModel=LLaVA1.5, Decoding Strategy=HGAI2025.08 | 90.7 | 90.77 | 87.67 | 88.12 | 81 | 82.8 | — | — | — | — | — | — | — | — | |
| MAPModel=LLaVA1.5, Decoding Strategy=MAP2025.08 | 90.1 | 89.86 | 87.4 | 87.44 | 81.8 | 82.82 | — | — | — | — | — | — | — | — | |
| DAMOModel=LLaVA1.5, Decoding Strategy=DAMO2025.08 | 90.03 | 90.02 | 86.73 | 87.13 | 80.47 | 82.15 | — | — | — | — | — | — | — | — | |
| DCLAModel=LLaVA1.5, Decoding Strategy=DCLA2025.08 | 90.03 | 89.99 | 86.73 | 87.1 | 80.7 | 82.28 | — | — | — | — | — | — | — | — | |
| SPINModel=LLaVA1.5, Decoding Strategy=SPIN2025.08 | 89.63 | 89.67 | 86.43 | 86.9 | 79.9 | 81.74 | — | — | — | — | — | — | — | — | |
| VanillaModel=LLaVA1.5, Decoding Strategy=Vanilla2025.08 | 89.57 | 89.66 | 86.1 | 86.7 | 79.5 | 81.48 | — | — | — | — | — | — | — | — | |
| VCDModel=LLaVA1.5, Decoding Strategy=VCD2025.08 | 89.07 | 89 | 85.63 | 86.03 | 79.27 | 81.01 | — | — | — | — | — | — | — | — | |
| MAPModel=mPLUG2, Decoding Strategy=MAP2025.08 | 88.8 | 88.16 | 86.7 | 86.25 | 84.37 | 84.21 | — | — | — | — | — | — | — | — | |
| DCLAModel=mPLUG2, Decoding Strategy=DCLA2025.08 | 88.53 | 87.99 | 86.03 | 86.05 | 82.67 | 83.25 | — | — | — | — | — | — | — | — | |
| SPINModel=mPLUG2, Decoding Strategy=SPIN2025.08 | 88.47 | 87.71 | 86.63 | 86.03 | 84.37 | 84.04 | — | — | — | — | — | — | — | — | |
| HGAIModel=mPLUG2, Decoding Strategy=HGAI2025.08 | 88.43 | 87.67 | 86.47 | 85.87 | 84.3 | 83.97 | — | — | — | — | — | — | — | — | |
| DAMOModel=mPLUG2, Decoding Strategy=DAMO2025.08 | 88.4 | 87.75 | 86.4 | 85.92 | 83.9 | 83.75 | — | — | — | — | — | — | — | — | |
| DCLAModel=InstBLIP, Decoding Strategy=DCLA2025.08 | 88.4 | 87.57 | 83.97 | 82.44 | 82.77 | 81.37 | — | — | — | — | — | — | — | — | |
| MAPModel=InstBLIP, Decoding Strategy=MAP2025.08 | 88.3 | 87.43 | 85.43 | 84.82 | 82.97 | 82.14 | — | — | — | — | — | — | — | — | |
| VanillaModel=mPLUG2, Decoding Strategy=Vanilla2025.08 | 88.27 | 87.56 | 86.37 | 85.82 | 84.07 | 83.82 | — | — | — | — | — | — | — | — | |
| DAMOModel=InstBLIP, Decoding Strategy=DAMO2025.08 | 88.23 | 87.46 | 84.93 | 84.49 | 82.5 | 82.31 | — | — | — | — | — | — | — | — | |
| HGAIModel=InstBLIP, Decoding Strategy=HGAI2025.08 | 87.3 | 85.87 | 84.73 | 83.49 | 82.8 | 81.79 | — | — | — | — | — | — | — | — | |
| VanillaModel=InstBLIP, Decoding Strategy=Vanilla2025.08 | 87.13 | 85.72 | 83.5 | 84.09 | 80.77 | 81.93 | — | — | — | — | — | — | — | — | |
| VCDModel=mPLUG2, Decoding Strategy=VCD2025.08 | 86.77 | 86.11 | 83.97 | 83.59 | 82.1 | 82.07 | — | — | — | — | — | — | — | — | |
| SPINModel=InstBLIP, Decoding Strategy=SPIN2025.08 | 86.33 | 84.61 | 84.1 | 82.53 | 82.83 | 81.4 | — | — | — | — | — | — | — | — | |
| SDCDModel=LLaVA1.5, Decoding=SDCD2026.01 | 85.9 | 84.56 | 83.43 | 82.33 | 80.87 | 80.12 | — | — | 93.46 | 77.2 | 88.19 | 77.2 | 83.36 | 77.13 | |
| SDCDModel=Qwen2.5-VL, Decoding=SDCD2026.01 | 85.63 | 83.38 | 85.07 | 82.89 | 83.93 | 81.76 | — | — | 98.9 | 72.07 | 97.05 | 72.33 | 94.57 | 72 | |
| VCDModel=LLaVA1.5, Decoding=VCD2026.01 | 84.87 | 83.37 | 82.43 | 81.2 | 79.9 | 79.06 | — | — | 92.52 | 75.87 | 87.34 | 75.87 | 82.52 | 75.87 | |
| VCDModel=InstBLIP, Decoding Strategy=VCD2025.08 | 84.1 | 83.07 | 80.17 | 79.27 | 78.17 | 78.13 | — | — | — | — | — | — | — | — | |
| RegularModel=LLaVA1.5, Decoding=Regular2026.01 | 82.93 | 80.87 | 81.1 | 79.24 | 78.63 | 77.13 | — | — | 92.01 | 72.13 | 87.9 | 72.13 | 82.96 | 72.07 | |
| VCDModel=Qwen2.5-VL, Decoding=VCD2026.01 | 82.67 | 79.23 | 82.97 | 79.89 | 81.77 | 78.35 | — | — | 98.8 | 66.13 | 97.5 | 67.67 | 96.4 | 66 | |
| RegularModel=Qwen2.5-VL, Decoding=Regular2026.01 | 81.73 | 77.94 | 81.23 | 77.33 | 80.83 | 77.08 | — | — | 98.37 | 64.53 | 97.66 | 64 | 95.84 | 64.47 | |
| OSGAModel Architecture=LLaVA-v1.52026.01 | 0.8787 | 0.8643 | 0.8637 | 0.85 | 0.845 | 0.8329 | 0.8629 | 0.8491 | — | — | — | — | — | — | |
| OPERAModel Architecture=LLaVA-v1.52026.01 | 0.8753 | 0.8645 | 0.8421 | 0.835 | 0.8088 | 0.8069 | 0.8421 | 0.8355 | — | — | — | — | — | — | |
| VCDModel Architecture=LLaVA-v1.52026.01 | 0.8684 | 0.8683 | 0.8265 | 0.8337 | 0.7731 | 0.7928 | 0.8227 | 0.8316 | — | — | — | — | — | — | |
| ICDModel Architecture=LLaVA-v1.52026.01 | 0.8487 | 0.8327 | 0.8293 | 0.8145 | 0.8107 | 0.7996 | 0.8296 | 0.8156 | — | — | — | — | — | — | |
| BaselineModel Architecture=LLaVA-v1.52026.01 | 0.8349 | 0.8228 | 0.7998 | 0.7934 | 0.7603 | 0.7626 | 0.7983 | 0.7929 | — | — | — | — | — | — |