3D Visual Grounding on ARKitScenes (test)
74.2Unique Success RatePanoGrounder
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PanoGrounderGT object segmentation=true2025.12 | 74.2 | 48 | 53.5 | |
| BUTD-DETRGT object segmentation=true, GT semantic labels=true2025.12 | 66.3 | 30.6 | 36.1 | |
| MCLNGT object segmentation=true, GT semantic labels=true2025.12 | 61.2 | 30 | 35.3 | |
| 3D-VisTAGT object segmentation=true2025.12 | 59.7 | 26.6 | 32.9 | |
| ViL3DRelGT object segmentation=true2025.12 | 57.2 | 21.1 | 28.3 |