Object Hallucination Evaluation on POPE GQA Popular
89.4AccuracySECOND
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SECONDModel=LLaVA-Next (CLIP-336), LLM=Vicuna-7B, CD=true2025.06 | 89.4 | — | 87.4 | 85.5 | |
| SIRABackbone=Qwen2.5-VL2026.05 | 88.37 | — | 88.6 | — | |
| SECONDModel=LLaVA-OneVision (SigLIP-384), LLM=Qwen2-0.5B, CD=true2025.06 | 87.8 | — | 87.2 | 83.3 | |
| SECONDModel=LLaVA-OneVision (SigLIP-384), LLM=Qwen2-0.5B, CD=false2025.06 | 87.4 | — | 86.7 | 82.2 | |
| HDDModel=LLaVA-NeXT2025.12 | 86.8 | — | 87.3 | — | |
| baselineModel=LLaVA-OneVision (SigLIP-384), LLM=Qwen2-0.5B, CD=false2025.06 | 86.8 | — | 86.3 | 83.1 | |
| VCDModel=LLaVA-OneVision (SigLIP-384), LLM=Qwen2-0.5B, CD=true2025.06 | 86.7 | — | 86.4 | 84.3 | |
| baselineModel=LLaVA-Next (CLIP-336), LLM=Vicuna-7B, CD=false2025.06 | 86.6 | — | 86.3 | 84.8 | |
| SECONDModel=LLaVA-Next (CLIP-336), LLM=Vicuna-7B, CD=false2025.06 | 86.5 | — | 86.3 | 84.9 | |
| ICTBackbone=Qwen2.5-VL2026.05 | 86.43 | — | 85.47 | — | |
| MaskCDBackbone=Qwen2.5-VL2026.05 | 86.35 | — | 86.7 | — | |
| CMPCBase Model=LLaVA 1.52025.01 | 85.5 | 85.24 | 85.55 | 85.87 | |
| SECONDModel=LLaVA-Next (CLIP-336), LLM=Mistral-7B, CD=true2025.06 | 85.3 | — | 87.5 | 92.1 | |
| SECONDModel=LLaVA-Next (CLIP-336), LLM=Mistral-7B, CD=false2025.06 | 84.8 | — | 85.2 | 87.2 | |
| VCDModel=LLaVA-Next (CLIP-336), LLM=Vicuna-7B, CD=true2025.06 | 84.6 | — | 84.9 | 86.3 | |
| VCDModel=LLaVA-Next (CLIP-336), LLM=Mistral-7B, CD=true2025.06 | 84.5 | — | 85.1 | 88 | |
| RITUALModel=LLaVA-NeXT2025.12 | 84.4 | — | 86.2 | — | |
| baselineModel=LLaVA-Next (CLIP-336), LLM=Mistral-7B, CD=false2025.06 | 84.2 | — | 84.8 | 88.2 | |
| CoFi-DecBackbone=Qwen-VL2025.12 | 83.54 | 87.24 | 83.09 | — | |
| SIRABackbone=LLaVA-v1.52026.05 | 82.93 | — | 83.75 | — | |
| SECONDModel=Yi-VL (CLIP-448), LLM=Yi-6B, CD=true2025.06 | 82.8 | — | 82.9 | 83.3 | |
| DoLaBackbone=Qwen2.5-VL2026.05 | 82.53 | — | 82.35 | — | |
| SECONDModel=Yi-VL (CLIP-448), LLM=Yi-6B, CD=false2025.06 | 82.5 | — | 81.4 | 76.6 | |
| M3IDBackbone=Qwen-VL2025.12 | 82.13 | 84.58 | 81.48 | — | |
| DeGFBackbone=Qwen-VL2025.12 | 82.1 | 86.39 | 81.85 | — | |
| MaskCDBackbone=LLaVA-v1.52026.05 | 82.05 | — | 83.1 | — | |
| VCDBackbone=Qwen-VL2025.12 | 81.97 | 82.82 | 81.73 | — | |
| OPERABackbone=Qwen2.5-VL2026.05 | 81.97 | — | 82.12 | — | |
| VCDBackbone=Qwen2.5-VL2026.05 | 81.83 | — | 82.23 | — | |
| SIDBackbone=Qwen2.5-VL2026.05 | 81.61 | — | 81.83 | — | |
| SIDBackbone=LLaVA-v1.52026.05 | 81.36 | — | 82.62 | — | |
| RegularBackbone=Qwen-VL2025.12 | 81.33 | 83.38 | 80.74 | — | |
| RITUALBackbone=Qwen-VL2025.12 | 81.13 | 85.48 | 81.03 | — | |
| baselineModel=Yi-VL (CLIP-448), LLM=Yi-6B, CD=false2025.06 | 81 | — | 79.7 | 74.5 | |
| BaseBackbone=Qwen2.5-VL2026.05 | 80.87 | — | 80.33 | — | |
| ICTBackbone=LLaVA-v1.52026.05 | 80.86 | — | 82.64 | — | |
| VCDBackbone=LLaVA-v1.52026.05 | 80.73 | — | 82.24 | — | |
| VCDBase Model=LLaVA 1.52025.01 | 80.73 | 76.26 | 82.24 | 89.24 | |
| HDDModel=LLaVA-1.52025.12 | 80.5 | — | 81.3 | — | |
| VCDModel=LLaVA-NeXT2025.12 | 80.3 | — | 81.1 | — | |
| VCDModel=LLaVA-1.52025.12 | 80.1 | — | 81.6 | — | |
| VCDModel=Yi-VL (CLIP-448), LLM=Yi-6B, CD=true2025.06 | 79.9 | — | 79.5 | 78.2 | |
| ICDBase Model=LLaVA 1.52025.01 | 79.87 | 80.07 | 79.8 | 79.53 | |
| CMPCBase Model=InstructBLIP2025.01 | 79.87 | 77.28 | 80.78 | 84.6 | |
| RITUALModel=LLaVA-1.52025.12 | 79.5 | — | 79.8 | — | |
| OPERABackbone=LLaVA-v1.52026.05 | 79.27 | — | 82.11 | — | |
| ICDBase Model=InstructBLIP2025.01 | 78.8 | 75.15 | 80.87 | 87.53 | |
| VCDBase Model=InstructBLIP2025.01 | 78.57 | 74.62 | 80.17 | 86.61 | |
| CoFi-DecBackbone=LLaVA-1.52025.12 | 78.56 | 73.28 | 83.34 | — | |
| SamplingModel=LLaVA-NeXT2025.12 | 78.2 | — | 77.1 | — | |
| BaseBackbone=LLaVA-v1.52026.05 | 78.17 | — | 78.37 | — | |
| BaselineBase Model=LLaVA 1.52025.01 | 78.17 | 77.64 | 78.37 | 79.12 | |
| SamplingModel=LLaVA-1.52025.12 | 78.1 | — | 78.3 | — | |
| HDDModel=InstructBLIP2025.12 | 77 | — | 80.1 | — | |
| VCDModel=InstructBLIP2025.12 | 76.8 | — | 78.6 | — | |
| RITUALModel=InstructBLIP2025.12 | 76.7 | — | 78.7 | — | |
| CoFi-DecBackbone=InstructBLIP2025.12 | 76.18 | 73.17 | 78.65 | — | |
| VCDBackbone=InstructBLIP2025.12 | 76.13 | 71.1 | 78.68 | — | |
| DeGFBackbone=InstructBLIP2025.12 | 75.34 | 71.89 | 77.96 | — | |
| M3IDBackbone=InstructBLIP2025.12 | 75.17 | 69.94 | 78.04 | — | |
| DeGFBackbone=LLaVA-1.52025.12 | 75.12 | 71.56 | 80.98 | — | |
| RITUALBackbone=LLaVA-1.52025.12 | 74.8 | 67.5 | 79.15 | — | |
| RITUALBackbone=InstructBLIP2025.12 | 74.5 | 69.17 | 77.61 | — | |
| DoLaBackbone=LLaVA-v1.52026.05 | 74.03 | — | 78.75 | — | |
| M3IDBackbone=LLaVA-1.52025.12 | 73.87 | 66.7 | 78.49 | — | |
| BaselineBase Model=InstructBLIP2025.01 | 73.87 | 69.63 | 76.42 | 84.69 | |
| SamplingModel=InstructBLIP2025.12 | 73.5 | — | 76.1 | — | |
| RegularBackbone=LLaVA-1.52025.12 | 73.47 | 66.83 | 77.84 | — | |
| RegularBackbone=InstructBLIP2025.12 | 73.33 | 68.72 | 76.26 | — | |
| VCDBackbone=LLaVA-1.52025.12 | 72.37 | 65.27 | 77.58 | — |