Object Hallucination Evaluation on MSCOCO POPE
91.63Random AccuracyDAC
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DACModel=LLaVA-NeXT, Decoding strategy=greedy decoding2025.02 | 91.63 | 91.32 | 89.27 | 89.14 | 86 | 85.71 | — | — | |
| OPERAModel=LLaVA-NeXT, Decoding strategy=beam search2025.02 | 91.36 | 90.8 | 88.65 | 88.6 | 85.1 | 85.75 | — | — | |
| VCDModel=LLaVA-NeXT, Decoding strategy=greedy decoding2025.02 | 91.3 | 90.8 | 88.63 | 88.31 | 85.53 | 85.58 | — | — | |
| BaselineModel=LLaVA-NeXT, Decoding strategy=greedy decoding2025.02 | 91.27 | 90.76 | 88.6 | 88.27 | 85.5 | 85.54 | — | — | |
| VCDModel=Qwen3-VL-4B2026.04 | 91.24 | 90.81 | 89.76 | 89.14 | 88.6 | 88.05 | — | — | |
| HTDCModel=Qwen3-VL-4B2026.04 | 91.24 | 90.81 | 89.77 | 89.14 | 88.6 | 88.05 | — | — | |
| SIDModel=LLaVA-NeXT, Decoding strategy=greedy decoding2025.02 | 91.2 | 90.73 | 88.6 | 88.3 | 85.87 | 85.89 | — | — | |
| RegularModel=Qwen3-VL-4B2026.04 | 91.03 | 90.58 | 89.5 | 88.85 | 88.47 | 87.89 | — | — | |
| DACModel=LLaVA-1.5, Decoding strategy=greedy decoding2025.02 | 90.83 | 90.6 | 89.5 | 89.1 | 84.12 | 84.42 | — | — | |
| DoLaModel=Qwen3-VL-4B2026.04 | 90.58 | 90.4 | 88.87 | 88.62 | 87.6 | 87.59 | — | — | |
| CATCHVenue=ECCV’242026.05 | 90.43 | 90.13 | 87.07 | 86.56 | 83.17 | 83.18 | 86.89 | 86.62 | |
| HIREBase Model=LLaVA-1.5-7B, TFLOPs=10.622026.03 | 90.37 | 90.25 | 87.7 | 87.86 | 83.73 | 83.56 | 87.27 | 87.23 | |
| HIREBase Model=InstructBLIP, TFLOPs=1.252026.03 | 90.3 | 89.83 | 84.03 | 84.25 | 81.47 | 82.17 | 85.27 | 85.42 | |
| CRG2026.05 | 90.3 | 89.51 | 88.4 | 86.54 | 84.43 | 83.77 | 87.71 | 86.61 | |
| DAMOModel=Qwen3-VL-4B2026.04 | 90.17 | 89.57 | 89.03 | 88.19 | 88.33 | 87.53 | — | — | |
| ICTVenue=CVPR’252026.05 | 90.11 | 90.03 | 87.5 | 87.6 | 84.43 | 83.74 | 87.35 | 87.12 | |
| ICT2026.05 | 90.11 | 90.03 | 87.5 | 87.6 | 84.43 | 83.74 | 87.35 | 87.12 | |
| VDDVenue=arXiv’242026.05 | 90 | 88.79 | 85.91 | 84.4 | 83.52 | 82.2 | 86.48 | 85.13 | |
| OPERAModel=LLaVA-1.5, Decoding strategy=beam search2025.02 | 89.87 | 89.95 | 86.3 | 86.88 | 79.77 | 81.77 | — | — | |
| CASTBackbone=LLaVA-1.5-7b, α=1.5, K=100, Decoding Setting=direct sampling2026.05 | 89.87 | 89.43 | 88.32 | 87.95 | 84.27 | 84.41 | 87.49 | 87.22 | |
| CCAModel=LLaVA-1.5, Decoding strategy=greedy decoding2025.02 | 89.77 | 89.05 | 86.45 | 86.02 | 83.97 | 83.82 | — | — | |
| ONLY2026.05 | 89.7 | 89.1 | 86 | 86.31 | 79.4 | 81.07 | 85.03 | 85.49 | |
| BaselineModel=LLaVA-1.5, Decoding strategy=greedy decoding2025.02 | 89.63 | 89.74 | 86.23 | 86.82 | 79.7 | 81.71 | — | — | |
| ICDVenue=EMNLP’24 findings2026.05 | 89.56 | 89.68 | 86.16 | 86.76 | 79.71 | 81.7 | 85.14 | 86.05 | |
| VTIBase Model=LLaVA-1.5-7B2026.03 | 89.5 | 88.89 | 87.36 | 86.69 | 82.57 | 82.11 | 86.48 | 85.9 | |
| SIDModel=LLaVA-1.5, Decoding strategy=greedy decoding2025.02 | 89.43 | 89.08 | 85.93 | 85.94 | 80.43 | 81.47 | — | — | |
| IMCCDVenue=arXiv’252026.05 | 89.23 | 88.68 | 86.73 | 86.13 | 82.87 | 82.77 | 86.27 | 85.86 | |
| OPERAVenue=CVPR’242026.05 | 89.2 | 88.81 | 86.64 | 86.62 | 81.24 | 81.38 | 85.7 | 85.6 | |
| CAUSALMMVenue=ICLR’252026.05 | 88.93 | 88.1 | 87.13 | 87.26 | 83.7 | 82.78 | 86.59 | 86.05 | |
| CAUSALMM2026.05 | 88.93 | 88.1 | 87.13 | 87.26 | 83.7 | 82.78 | 86.59 | 86.05 | |
| CHASDModel=LLaVA-1.5-13B2026.05 | 88.89 | 87.97 | 87.03 | 86.75 | 84.91 | 84.35 | — | — | |
| HACLBase Model=LLaVA-1.5-7B2026.03 | 88.59 | 88.7 | 87.84 | 87.36 | 86.54 | 85.73 | 87.66 | 87.26 | |
| AvisCModel=LLaVA-1.5-13B2026.05 | 88.4 | 88.77 | 85.73 | 86.53 | 79.27 | 81.47 | — | — | |
| HTDCModel=LLaVA-v1.52026.04 | 88.25 | 87.42 | 87.3 | 86.18 | 85.23 | 84.29 | — | — | |
| DoLaModel=LLaVA-v1.52026.04 | 88.21 | 87.37 | 87.9 | 87.36 | 84.03 | 83.96 | — | — | |
| DAMROVenue=EMNLP’242026.05 | 88.2 | 87.29 | 85.67 | 84.98 | 82.07 | 81.9 | 85.31 | 84.72 | |
| VCDModel=LLaVA-v1.52026.04 | 88.18 | 87.34 | 87.27 | 86.14 | 85.13 | 84.18 | — | — | |
| RegularModel=LLaVA-v1.52026.04 | 88.04 | 87.18 | 88.56 | 87.56 | 85.07 | 84.13 | — | — | |
| DACModel=mPLUG-Owl2, Decoding strategy=greedy decoding2025.02 | 87.71 | 87.57 | 84.96 | 84.46 | 82.58 | 82.32 | — | — | |
| WoodpeckerVenue=SCIS’242026.05 | 87.67 | 86.45 | 80.67 | 79.72 | 80.67 | 80 | 83 | 82.05 | |
| VCDModel=LLaVA-1.5, Decoding strategy=greedy decoding2025.02 | 87.53 | 87.81 | 84.43 | 85.2 | 78.13 | 80.38 | — | — | |
| ICDBase Model=LLaVA-1.5-7B2026.03 | 87.51 | 83.28 | 83.15 | 83.91 | 79.13 | 80.41 | 83.26 | 82.53 | |
| OctopusBase Model=LLaVA-1.5-7B, TFLOPs=16.342026.03 | 87.51 | 85.4 | 85.2 | 84.19 | 82.22 | 81.44 | 85.79 | 83.44 | |
| ICDModel=Qwen3-VL-4B2026.04 | 87.33 | 86.15 | 88.1 | 86.87 | 88.66 | 87.75 | — | — | |
| DAMOModel=LLaVA-v1.52026.04 | 86.87 | 85.64 | 86.03 | 84.46 | 84.33 | 82.9 | — | — | |
| OctopusBase Model=InstructBLIP, TFLOPs=2.272026.03 | 86.63 | 85.3 | 84.9 | 83.55 | 82.83 | 81.43 | 84.79 | 83.43 | |
| VCDModel=LLaVA-1.5-13B2026.05 | 86.63 | 86.45 | 83.42 | 83.73 | 79.69 | 80.66 | — | — | |
| CHASDModel=Qwen2.5-VL2026.05 | 86.37 | 84.31 | 85.57 | 83.54 | 84.96 | 82.97 | — | — | |
| SIDModel=mPLUG-Owl2, Decoding strategy=greedy decoding2025.02 | 86.3 | 86.86 | 81.27 | 82.82 | 77.27 | 79.89 | — | — | |
| BaselineModel=mPLUG-Owl2, Decoding strategy=greedy decoding2025.02 | 86.27 | 86.88 | 80.73 | 82.52 | 76.17 | 77.69 | — | — | |
| OPERAModel=mPLUG-Owl2, Decoding strategy=beam search2025.02 | 86.23 | 86.84 | 80.7 | 82.48 | 76.87 | 78.01 | — | — | |
| M3IDVenue=CVPR’242026.05 | 86.2 | 84.51 | 84.77 | 83.17 | 82.53 | 81.14 | 84.5 | 82.94 | |
| M3IDBase Model=LLaVA-1.5-7B, TFLOPs=16.262026.03 | 86.13 | 81.85 | 82.07 | 80.77 | 79.5 | 78.15 | 82.57 | 80.26 | |
| AVISCBase Model=InstructBLIP, TFLOPs=2.162026.03 | 86.03 | 84.41 | 84.27 | 82.77 | 81.83 | 80.67 | 84.04 | 82.62 | |
| AvisCModel=Qwen2.5-VL2026.05 | 85.63 | 83.43 | 84.57 | 82.36 | 83.9 | 82.03 | — | — | |
| VCDBase Model=LLaVA-1.5-7B, TFLOPs=16.262026.03 | 85.43 | 83.99 | 83.17 | 81.94 | 80.27 | 79.49 | 82.96 | 81.81 | |
| ConVisBase Model=LLaVA-1.5-7B2026.03 | 84.7 | — | 83.2 | — | 81.1 | — | 83 | — | |
| AVISCBase Model=LLaVA-1.5-7B, TFLOPs=16.262026.03 | 84.67 | 82.21 | 83.67 | 81.27 | 81.83 | 79.55 | 83.39 | 81.01 | |
| OPERABase Model=InstructBLIP2026.03 | 84.57 | 83.74 | 78.24 | 79.15 | 74.59 | 76.33 | 79.13 | 79.74 | |
| VCDModel=mPLUG-Owl2, Decoding strategy=greedy decoding2025.02 | 84.4 | 84.79 | 81 | 81.12 | 77.1 | 77 | — | — | |
| OPERABase Model=LLaVA-1.5-7B2026.03 | 84.4 | — | 83.4 | — | 81.2 | — | 83 | — | |
| ICDBase Model=InstructBLIP2026.03 | 84.36 | 83.82 | 77.88 | 78.7 | 75.17 | 77.23 | 79.14 | 79.92 | |
| ICDModel=LLaVA-v1.52026.04 | 84.33 | 83.3 | 86.07 | 84.87 | 84.33 | 83.3 | — | — | |
| LLaVA-1.5-7BBase Model=LLaVA-1.5-7B, TFLOPs=8.132026.03 | 83.77 | 81.94 | 82.57 | 80.86 | 79.77 | 78.47 | 82.04 | 80.42 | |
| RegularBackbone=LLaVA-1.5-7b, Decoding Setting=direct sampling2026.05 | 83.29 | 81.33 | 81.88 | 80.06 | 78.96 | 77.57 | 81.38 | 79.65 | |
| VCDModel=Qwen2.5-VL2026.05 | 82.67 | 79.23 | 82.97 | 79.89 | 81.77 | 78.35 | — | — | |
| M3IDBase Model=InstructBLIP, TFLOPs=2.162026.03 | 82.33 | 81.53 | 80.9 | 80.42 | 78.53 | 78.49 | 80.59 | 80.15 | |
| VCDBase Model=InstructBLIP, TFLOPs=2.162026.03 | 82.03 | 81.56 | 79.13 | 79.2 | 77.23 | 77.72 | 79.46 | 79.49 | |
| GreedyModel=Qwen2.5-VL2026.05 | 81.73 | 77.94 | 81.23 | 77.33 | 80.83 | 77.08 | — | — | |
| InstructBLIPBase Model=InstructBLIP, TFLOPs=1.082026.03 | 81.53 | 81.19 | 78.47 | 78.75 | 77.43 | 78 | 79.14 | 79.31 | |
| GreedyModel=LLaVA-1.5-13B2026.05 | 79.33 | 79.63 | 76.3 | 77.32 | 73.44 | 75.12 | — | — | |
| AVISCBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 84.86 | 83.55 | |
| AVISCBase Model=InstructBLIP2026.05 | — | — | — | — | — | — | 83.38 | 82.29 | |
| DMAS2026.05 | — | — | — | — | — | — | 86.81 | 86.79 | |
| EVASBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 84.25 | 82.86 | |
| ICDBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 83.37 | 82.37 | |
| InstructBLIPBase Model=InstructBLIP2026.05 | — | — | — | — | — | — | 80.37 | 80.5 | |
| LLaVA-1.5-7BBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 83.93 | 82.24 | |
| M3IDBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 82.46 | 80.72 | |
| VCDBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 84.68 | 83.61 | |
| VCDBase Model=InstructBLIP2026.05 | — | — | — | — | — | — | 83.49 | 82.68 | |
| YARDBase Model=LLaVA-1.5-7B2026.05 | — | — | — | — | — | — | 86.14 | 85.85 | |
| YARDBase Model=InstructBLIP2026.05 | — | — | — | — | — | — | 89.06 | 88.45 |