Scene Understanding on OmniDrive (test)
0.341ROUGE-LGaussianDWM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GaussianDWMReference=-, # LLM Params=8B, Category=Unified Understanding and Generation2025.12 | 0.341 | 0.395 | 0.712 | |
| Separated unificationReference=-, # LLM Params=1.8B, Category=Unified Understanding and Generation2025.12 | 0.327 | 0.384 | 0.745 | |
| HERMESReference=ICCV 25, # LLM Params=1.8B, Category=Unified Understanding and Generation2025.12 | 0.327 | 0.384 | 0.741 | |
| OmniDriveReference=CVPR 25, # LLM Params=7B, Category=Only Understanding2025.12 | 0.326 | 0.38 | 0.686 | |
| OmniDrive-2DReference=CVPR 25, # LLM Params=7B, Category=Only Understanding2025.12 | 0.325 | 0.383 | 0.671 | |
| OmniDrive-BEVReference=CVPR 25, # LLM Params=7B, Category=Only Understanding2025.12 | 0.278 | 0.356 | 0.595 | |
| GPT-4oReference=-, # LLM Params=-, Category=Only Understanding2025.12 | 0.223 | — | 0.244 | |
| LLaVA-OVReference=arXiv 24, # LLM Params=7B, Category=Only Understanding2025.12 | 0.221 | — | 0.284 |