3D visual grounding on ScanRefer Box-Level
55.5Accuracy @ IoU 0.25Chat-Scene
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Chat-SceneModel Category=3D LMMs2025.01 | 55.5 | 50.2 | |
| 3D-LLaVAModel Category=3D LMMs2025.01 | 51.2 | 40.6 | |
| 3D-VisTAModel Category=Specialist Models2025.01 | 50.6 | 45.8 | |
| ConcretNetModel Category=Specialist Models2025.01 | 50.6 | 46.5 | |
| Grounded 3D-LLMModel Category=3D LMMs2025.01 | 48.6 | 44 | |
| 3D-LLMModel Category=3D LMMs2025.01 | 30.3 | — |