Video 3D Perception on ARKitScenes
15.6F1 Score @ IoU=0.25Spatial Encoder
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Spatial EncoderInput=Video detector2026.03 | 15.6 | 8.2 | |
| SpatialLMInput=Point-cloud detector2026.03 | 12.2 | 5.5 | |
| DetAny3DInput=Image detector2026.03 | 9.4 | 1.2 | |
| 3D-MOODInput=Image detector2026.03 | 6.6 | 0.9 | |
| Qwen3-VL-4BInput=MLLMs2026.03 | 4.1 | 1.1 | |
| Qwen3-VL-235BInput=MLLMs2026.03 | 3.4 | 0.8 | |
| SceneScriptInput=Point-cloud detector2026.03 | 2 | 1.4 |