Motion Understanding on AV-Car
91.5AccuracyNVILA-Video-15B
Evaluation Results
| Method | Links | |
|---|---|---|
| NVILA-Video-15BMode=Fine-tuned, Data=FoundationMotion2025.12 | 91.5 | |
| NVILA-Video-8BMode=Fine-tuned, Data=FoundationMotion2025.12 | 90.6 | |
| NVILA-Video-8BMode=Zero-shot2025.12 | 88.9 | |
| NVILA-Video-8BMode=Fine-tuned, Data=PLM 467k2025.12 | 87.9 | |
| NVILA-Video-15BMode=Zero-shot2025.12 | 84.4 | |
| Gemini-2.5-FlashMode=Zero-shot2025.12 | 84.1 | |
| Qwen-2.5-VL-72BMode=Zero-shot2025.12 | 83.3 | |
| Qwen-2.5-VL-7BMode=Fine-tuned, Data=FoundationMotion2025.12 | 82.1 | |
| Qwen-2.5-VL-7BMode=Zero-shot2025.12 | 80.3 | |
| NVILA-Video-15BMode=Fine-tuned, Data=PLM 467k2025.12 | 79.4 |