Object Hallucination Evaluation on POPE Popular LLaVA-1.5
88POPE Score (Zh)CLAIM
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| CLAIMBase Model=LLaVA-1.52025.06 | 88 | 87.53 | 84.03 | 86.13 | 80.07 | 85.15 | — | — | — | |
| CLAIMBackbone=LLaVA-1.52025.06 | 87.53 | 84.03 | 86.13 | 80.07 | — | 85.15 | 88 | — | — | |
| CLAIMBackbone=Qwen-VL-Chat2025.06 | 85.9 | 80.63 | 69.23 | — | — | 81.53 | 85.47 | 86.43 | — | |
| PAIBase Model=LLaVA-1.52025.06 | 83.9 | 67.73 | 79.17 | 82.9 | 71.9 | 77.12 | — | — | — | |
| BaselineBackbone=Qwen-VL-Chat2025.06 | 83.4 | 68.03 | 75.73 | — | — | 73.31 | 85.67 | 62.2 | 77.2 | |
| VCDBase Model=LLaVA-1.52025.06 | 83.2 | 67.2 | 70.17 | 80.1 | 69.47 | 74.03 | — | — | — | |
| VCDBackbone=LLaVA-1.52025.06 | 83.2 | 67.2 | 70.17 | 80.1 | 69.47 | 74.03 | — | — | — | |
| BaselineBase Model=LLaVA-1.52025.06 | 83.07 | 62.93 | 69.2 | 82.03 | 71.17 | 73.68 | — | — | — | |
| BaselineBackbone=LLaVA-1.52025.06 | 83.07 | 62.93 | 69.2 | 82.03 | 71.17 | 73.68 | 87.43 | — | — | |
| VCDBackbone=Qwen-VL-Chat2025.06 | 82.7 | 72.7 | 74.2 | — | — | 72.74 | — | 57.4 | 76.7 |