Embodied spatial reasoning on UrbanVideo-Bench
57.3CleanRobust-TO (Qwen3-VL-7B-Instruct)
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Robust-TO (Qwen3-VL-7B-Instruct)Frames=20.72026.06 | 57.3 | 54.1 | 54 | 54.9 | 53.5 | 55.1 | 54.3 | |
| Video-R1 (Qwen3-VL-7B-Instruct)Frames=322026.06 | 52 | 48.5 | 48 | 49 | 47.5 | 49.5 | 48.5 | |
| Robust-TO (Qwen2.5-VL-7B-Instruct)Frames=20.72026.06 | 50.6 | 47 | 46.5 | 47.7 | 46.1 | 48.3 | 47.1 | |
| Gemini-2.5-ProFrames=1fps2026.06 | 44.3 | 37.7 | 37.7 | 38.8 | 36.4 | 39.8 | 38.1 | |
| Video-R1 (Qwen2.5-VL-7B-Instruct)Frames=322026.06 | 43 | 38.5 | 38 | 39.5 | 37 | 40.5 | 38.7 | |
| Qwen2.5-VL-7B-Instruct (SFT)Frames=322026.06 | 39 | 30.9 | 30.7 | 32.1 | 29.2 | 33.3 | 31.2 | |
| GPT-4oFrames=322026.06 | 37.4 | 32.2 | 31.7 | 32.5 | 30.8 | 33.6 | 32.2 | |
| Qwen2.5-VL-72B-InstructFrames=322026.06 | 33.4 | 26.3 | 26 | 27.2 | 24.3 | 28.7 | 26.5 | |
| Qwen2.5-VL-7B-InstructFrames=322026.06 | 26.9 | 17 | 16.6 | 18.9 | 14.6 | 20.3 | 17.5 |