Fine-grained Video Understanding on MVBench
73.8AccuracyPVC InternVL2
Evaluation results
| Method | Links | |
|---|---|---|
| PVC InternVL2Size=8B, # token/frame=642024.12 | 73.8 | |
| PVC InternVL2Size=2B, # token/frame=642024.12 | 69.4 | |
| IXC-2.5Size=7B, # token/frame=4002024.12 | 69.1 | |
| InternVideo2Size=8B, # token/frame=962024.12 | 67.2 | |
| Qwen2-VLSize=7B, # token/frame=*2024.12 | 67 | |
| LongVUSize=7B, # token/frame=64-1442024.12 | 66.9 | |
| InternVL2Size=8B, # token/frame=2562024.12 | 66.4 | |
| Video-CCAMSize=9B, # token/frame=64/10.72024.12 | 64.6 | |
| Oryx MLLMSize=7B, # token/frame=*2024.12 | 63.9 | |
| Qwen2-VLSize=2B, # token/frame=*2024.12 | 63.2 | |
| Video-CCAMSize=4B, # token/frame=64/10.72024.12 | 62.8 | |
| TarsierSize=7B, # token/frame=5762024.12 | 62.6 | |
| KangarooSize=8B, # token/frame=10242024.12 | 61.1 | |
| VideoChat2Size=7B, # token/frame=962024.12 | 60.4 | |
| InternVL2Size=2B, # token/frame=2562024.12 | 60.2 | |
| LLaVA-VideoSize=7B, # token/frame=1692024.12 | 58.6 | |
| LLaVA-OVSize=7B, # token/frame=1962024.12 | 56.7 | |
| VideoLLaMA2Size=7B, # token/frame=722024.12 | 54.6 | |
| Gemini-1.5-Pro2024.12 | 52.6 | |
| GPT-4o2024.12 | 47.8 | |
| PLLAVASize=7B, # token/frame=1442024.12 | 46.6 | |
| GPT-4V2024.12 | 43.7 |