Hallucination Detection on EventHallusion Entire, Mix, Misleading
69.3Accuracy (Entire)DINO-HEAL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DINO-HEALBase Model=Qwen2.5-VL-7B2026.04 | 69.3 | 58.55 | 86.27 | 68.46 | |
| DTRBase Model=Qwen2.5-VL-7B2026.04 | 65.79 | 69.95 | 90.2 | 73.84 | |
| BaselineBase Model=Qwen2.5-VL-7B2026.04 | 64.04 | 60.1 | 88.24 | 68.22 | |
| TCDBase Model=Qwen3-VL-8B2026.04 | 63.16 | 65.8 | 97.06 | 72.86 | |
| DTRBase Model=Qwen3-VL-8B2026.04 | 63.16 | 72.54 | 96.08 | 75.79 | |
| BaselineBase Model=Qwen3-VL-8B2026.04 | 62.28 | 65.28 | 96.08 | 72.13 | |
| DINO-HEALBase Model=Qwen3-VL-8B2026.04 | 62.28 | 66.84 | 96.08 | 72.86 | |
| TCDBase Model=Qwen2.5-VL-7B2026.04 | 61.4 | 63.73 | 84.31 | 68.22 | |
| VCDBase Model=Qwen3-VL-8B2026.04 | 60.53 | 62.69 | 97.06 | 70.66 | |
| VCDBase Model=Qwen2.5-VL-7B2026.04 | 59.65 | 60.1 | 86.27 | 66.5 | |
| LLaVA-NeXT-Video-7B + DTRNumber of frames=162026.04 | 54.39 | 62.18 | 70.59 | 62.1 | |
| DTRModel=Video-LLaVA-7B2026.04 | 52.63 | 46.63 | 77.45 | 55.99 | |
| DTRModel=LLaVA-NeXT-Video-7B2026.04 | 51.75 | 64.25 | 66.67 | 61.37 | |
| DINO-HEALModel=LLaVA-NeXT-Video-7B2026.04 | 47.37 | 67.88 | 50.98 | 57.95 | |
| LLaVA-NeXT-Video-7BNumber of frames=162026.04 | 46.49 | 68.91 | 59.8 | 60.39 | |
| TCDModel=LLaVA-NeXT-Video-7B2026.04 | 45.61 | 69.43 | 52.94 | 58.68 | |
| BaselineModel=LLaVA-NeXT-Video-7B2026.04 | 44.74 | 68.39 | 50 | 57.21 | |
| VCDModel=LLaVA-NeXT-Video-7B2026.04 | 42.11 | 63.73 | 51.96 | 54.77 | |
| VCDModel=Video-LLaVA-7B2026.04 | 40.35 | 52.85 | 39.22 | 45.97 | |
| TCDModel=Video-LLaVA-7B2026.04 | 31.58 | 68.91 | 40.2 | 51.34 | |
| DINO-HEALModel=Video-LLaVA-7B2026.04 | 29.82 | 57.51 | 42.16 | 45.97 | |
| BaselineModel=Video-LLaVA-7B2026.04 | 28.95 | 58.03 | 41.18 | 45.72 |