Video Hallucination Evaluation on Vript HAL
48.21RecallMultiToP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MultiToPBackbone=Qwen3-VL-4B-Instruct2026.06 | 48.21 | 70.68 | 56.58 | |
| TA-AEBackbone=Qwen3-VL-4B-Instruct2026.06 | 29.11 | 61.77 | 39.09 | |
| SmartSightBackbone=Qwen3-VL-4B-Instruct2026.06 | 28.51 | 64.51 | 39.32 | |
| Qwen3-VL-4B-InstructBackbone=Qwen3-VL-4B-Instruct2026.06 | 27 | 62.97 | 37.57 | |
| MultiToPBackbone=Video-LLaVA-7B2026.06 | 26.62 | 68.28 | 37.95 | |
| DINO-HEALBackbone=Qwen3-VL-4B-Instruct2026.06 | 26.54 | 63.73 | 37.1 | |
| MotionCDBackbone=Video-LLaVA-7B2026.06 | 25.94 | 55.26 | 35.31 | |
| TA-AEBackbone=Video-LLaVA-7B2026.06 | 25.94 | 59.74 | 36.17 | |
| TCDBackbone=Video-LLaVA-7B2026.06 | 25.82 | 54.49 | 35.04 | |
| DINO-HEALBackbone=Video-LLaVA-7B2026.06 | 25.82 | 52.44 | 34.6 | |
| SmartSightBackbone=Video-LLaVA-7B2026.06 | 25.75 | 58.67 | 35.79 | |
| MotionCDBackbone=Qwen3-VL-4B-Instruct2026.06 | 25.51 | 64.42 | 36.17 | |
| Video-LLaVA-7BBackbone=Video-LLaVA-7B2026.06 | 25.45 | 54 | 34.6 | |
| TCDBackbone=Qwen3-VL-4B-Instruct2026.06 | 25.33 | 63.98 | 35.92 |