Object Hallucination Detection on POPE
89.1AccuracyQwen2-VL-7B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen2-VL-7BModel=Qwen2-VL-7B*2026.05 | 89.1 | 94.9 | 82.7 | 88.4 | |
| PStarBackbone=Qwen2-VL-2B2026.05 | 88.7 | 98.1 | 78.8 | 88.5 | |
| PStarBackbone=Qwen2-VL-7B2026.05 | 88.7 | 99.9 | 77.4 | 88.7 | |
| Qwen2-VL-2BModel=Qwen2-VL-2B*2026.05 | 88.2 | 94.7 | 81 | 83.7 | |
| LLaMA-3.2-11B-VisionModel=LLaMA-3.2-11B-Vision*2026.05 | 88.1 | 88 | 88.3 | 88.1 | |
| Qwen2.5-VL-7BModel=Qwen2.5-VL-7B*2026.05 | 87.4 | 96.8 | 77.3 | 85.9 | |
| PStarBackbone=Qwen2.5-VL-7B2026.05 | 87.3 | 97.7 | 76.3 | 87.1 | |
| LLaVA-v1.5-7BModel=LLaVA-v1.5-7B*2026.05 | 87 | 92.1 | 80.9 | 86.1 | |
| InternVL2-2BModel=InternVL2-2B*2026.05 | 86.7 | 95.7 | 76.8 | 85.2 | |
| InternVL2-8BModel=InternVL2-8B*2026.05 | 86 | 96.6 | 74.7 | 84.2 | |
| GPT-4V (0409)Model=GPT-4V (0409)*2026.05 | 83.9 | 93.9 | 72.5 | 81.8 |