Video Understanding and Reasoning on Video-MME (test)
60.2Overall AccuracySmartSight
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SmartSightBase Model=Video-R12025.12 | 60.2 | 51.1 | 59.1 | 70.4 | |
| Video-R1Decoding Method=Greedy2025.12 | 59.8 | 50.8 | 59.7 | 69 | |
| VCDBase Model=Video-R12025.12 | 59.8 | 50.2 | 59.9 | 69.4 | |
| TCDBase Model=Video-R12025.12 | 59.7 | 50.1 | 59.1 | 70 | |
| DINO-HEALBase Model=Video-R12025.12 | 59.2 | 50.1 | 58.6 | 69 | |
| SmartSightBase Model=Qwen2.5-VL-7B2025.12 | 56.2 | 49.3 | 53.2 | 66.1 | |
| Qwen2.5-VL-7BDecoding Method=Greedy2025.12 | 54.8 | 45.4 | 52.5 | 66.5 | |
| VCDBase Model=Qwen2.5-VL-7B2025.12 | 54.8 | 45.1 | 52.3 | 67 | |
| TCDBase Model=Qwen2.5-VL-7B2025.12 | 54.7 | 44.1 | 53.1 | 66.8 | |
| DINO-HEALBase Model=Qwen2.5-VL-7B2025.12 | 54.6 | 47.4 | 51.9 | 64.5 | |
| SmartSightBase Model=LLaVA-NEXT-Video-7B2025.12 | 34 | 31.2 | 34.8 | 35.9 | |
| TCDBase Model=LLaVA-NEXT-Video-7B2025.12 | 31.1 | 27.9 | 30.9 | 34.7 | |
| VCDBase Model=LLaVA-NEXT-Video-7B2025.12 | 30.6 | 27.6 | 30.6 | 33.6 | |
| LLaVA-NEXT-Video-7BDecoding Method=Greedy2025.12 | 28.8 | 25.9 | 32.1 | 28.4 | |
| DINO-HEALBase Model=LLaVA-NEXT-Video-7B2025.12 | 28.3 | 28.1 | 28.1 | 28.6 |